Information processing device, information processing method, and program

Abstract

To predict an aging period when a circuit breaker malfunctions.SOLUTION: An information processing device having a processor and a storage device to predict an aging period when a circuit breaker including a trip coil for allowing a control current to flow to be excited and a contact for blocking current flowing in a power system when the trip coil is excited malfunctions stores a learning model for performing learning on the basis of learning data obtained by associating feature amounts showing a first waveform being a waveform of the control current acquired in each first aging years, and a second waveform being a waveform of the control current acquired in each second aging years different from the first aging years in the past with labels showing a first aging period when the first waveform cannot be acquired and a second aging period when the second waveform cannot be acquired, and currently predicts a third aging period when a third waveform cannot be acquired as an ageing period when the circuit breaker malfunctions by inputting the third waveform being a waveform of the control current.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing method, and a program for predicting when a circuit breaker will malfunction. [Background technology]

[0002] A circuit breaker is a switch that, in cooperation with a protective relay, cuts off the fault current flowing through the power system when an accident such as a short circuit or a ground fault occurs in the power system, thereby preventing the accident from spreading to the upstream power source and downstream load. For example, circuit breakers are installed at the power receiving inlet of a substation, the outlet of a transmission and distribution line, the primary or secondary side of a transformer, etc. Circuit breakers must cut off the fault current flowing through the power system. For each circuit breaker installed in a power system, maintenance is carried out on the circuit breaker's mechanism and control unit at regular intervals (e.g., every six years) after the circuit breaker is installed. (See, for example, Patent Document 1.) [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2021-103099 A Summary of the Invention [Problem to be solved by the invention]

[0004] Grease is applied to the circuit breaker mechanism so that the contacts that interrupt the fault current can be opened and closed smoothly. This grease deteriorates over time, drying, hardening, solidifying, etc. When the grease deteriorates, the operating speed of the plunger and latch of the circuit breaker mechanism slows down, which increases the possibility of the circuit breaker mechanism malfunctioning, and the control unit of the circuit breaker is more likely to malfunction due to the malfunction of the circuit breaker mechanism. The degree of deterioration of the grease varies depending on the surrounding environment in which the circuit breaker is installed and the number of times the circuit breaker has operated in the past. Therefore, it is difficult to predict the time when the circuit breaker will malfunction, and in the method of performing maintenance on the circuit breaker at regular intervals as described above, the circuit breaker may already be malfunctioning before the next maintenance of the circuit breaker. Even if an accident occurs in the power system at this timing, the circuit breaker will not be able to interrupt the fault current, and there is a risk that the effects of the accident will spread to the power source side and the load side.

[0005] The present invention has been made in consideration of the above-mentioned problems, and one object of the present invention is to provide an information processing device, an information processing method, and a program that are capable of predicting the time when a circuit breaker will malfunction, and enabling maintenance of the circuit breaker before the predicted time. [Means for solving the problem]

[0006] One of the present inventions for achieving the above-mentioned object is an information processing device having a processor and a storage device, which predicts a time point at which a circuit breaker including a trip coil that is excited when a control current flows through it and a contact that interrupts a current flowing through a power system when the trip coil is excited will malfunction, the information processing device storing a learning model that learns based on learning data that associates a first waveform, which is a waveform of the control current acquired for a first number of years in the past, and a second waveform, which is a waveform of the control current acquired for a second number of years of age different from the first number of years of age, with a label indicating the first aging time point at which the first waveform could not be acquired and the second aging time point at which the second waveform could not be acquired, and currently inputting a third waveform, which is the waveform of the control current, into the learning model, thereby predicting the third aging time point at which the third waveform cannot be acquired as the aging time at which the circuit breaker will malfunction.

[0007] According to the information processing device of the present invention, maintenance of the circuit breaker can be performed before the predicted aging time when the circuit breaker will malfunction, and therefore it is possible to interrupt a fault current flowing through the power system.

[0008] Another aspect of the present invention for achieving the above-mentioned object is an information processing device, which currently inputs the third waveform into the learning model to predict the time over time when the circuit breaker will malfunction, as well as the factors that will cause the circuit breaker to malfunction.

[0009] According to the information processing device of the present invention, it is possible to carry out maintenance focusing on factors that are predicted to cause the circuit breaker to malfunction before the time when the circuit breaker is predicted to malfunction, and therefore it is also possible to interrupt a fault current flowing through the power system.

[0010] Another aspect of the present invention for achieving the above-mentioned object is an information processing device, wherein the circuit breaker is a plurality of circuit breakers of the same specifications, the first waveform is the waveform of the control current in each of the plurality of circuit breakers of the same specifications, and the second waveform is the waveform of the control current in each of the plurality of circuit breakers of the same specifications.

[0011] According to the information processing device of the present invention, the learning data is data in which a first aging period and a second aging period are associated with each of a first waveform and a second waveform obtained from multiple circuit breakers of the same specifications. Therefore, by increasing the amount of learning data that the learning model learns, it becomes possible to accurately predict the aging period at which a circuit breaker will malfunction.

[0012] Another aspect of the present invention for achieving the above object is an information processing device, wherein the circuit breaker is a plurality of circuit breakers installed in a plurality of regions, the first waveform is the waveform of the control current in each of the plurality of circuit breakers, and the second waveform is the waveform of the control current in each of the plurality of circuit breakers.

[0013] According to the information processing device of the present invention, the learning data includes information on the area in which multiple circuit breakers are installed, and therefore the amount of learning data that the learning model learns from is further increased, making it possible to accurately predict the time over time when a circuit breaker will malfunction.

[0014] Other problems and solutions disclosed in the present application will be made clear in the section of the preferred embodiment and the drawings. Effect of the Invention

[0015] According to the present invention, by predicting the time when a circuit breaker will malfunction, it becomes possible to perform maintenance of the circuit breaker before the predicted time of aging, and therefore it becomes possible to interrupt a fault current flowing through a power system. [Brief description of the drawings]

[0016] [Figure 1] 2 is an equivalent circuit diagram showing an example of the relationship between a contact 110 and a tripping control circuit 120 included in the breaker 100. FIG. [Diagram 2] 11 is a graph showing a change in a control current flowing through a trip coil 123 while the trip coil 123 is excited. [Diagram 3]A block diagram showing a schematic configuration of a prediction system 200 including an information processing device 300 that predicts the aging time when a circuit breaker 100 will malfunction and the cause of the circuit breaker 100 malfunctioning using a learning model 343. [Figure 4A] FIG. 13 is a diagram showing an example of waveform data included in a first feature of the first learning data 342A. [Figure 4B] FIG. 13 is a diagram showing an example of waveform data included in a second feature of second learning data 342B. [Diagram 5] FIG. 13 is a diagram showing the structure of a deep neural network, which is an example of a learning model 343. [Figure 6] FIG. 2 is a block diagram showing an example of hardware of an information processing device 300 used to realize the prediction system 200. [Figure 7] 13 is a flowchart showing a process performed by the information processing device 300 to learn the learning model 343 using the first learning data 342A and the second learning data 342B. [Figure 8] 13 is a flowchart showing a process in which the information processing device 300 uses the learning model 343 to predict the current aging time when the circuit breaker 100 will malfunction and the cause of the circuit breaker 100 malfunctioning at that aging time. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] At least the following matters will become apparent from the description of this specification and the accompanying drawings. Hereinafter, the present invention will be described in detail based on an embodiment thereof with reference to the accompanying drawings.

[0018] 1 is an equivalent circuit diagram showing an example of the relationship between a contact 110 and a tripping control circuit 120 included in a circuit breaker 100. The circuit breaker 100 includes the contact 110 and the tripping control circuit 120 as components, and may be, for example, a vacuum circuit breaker, an oil circuit breaker, a porcelain circuit breaker, an air circuit breaker, or a gas circuit breaker. In this embodiment, the circuit breaker 100 is configured to interrupt a fault current that flows when an accident such as a short circuit or a ground fault occurs in an electric power transmission / distribution line.

[0019] The contact 110 is a switch that opens (trips) when a fault current flowing through the power transmission / distribution line is interrupted. The contact 110 includes a fixed electrode body 111 and a movable electrode body 112. One of the fixed electrode body 111 and the movable electrode body 112 is connected to one end of the power transmission / distribution line opposite to the power source side, and the other of the fixed electrode body 111 and the movable electrode body 112 is connected to one end of the power transmission / distribution line opposite to the load side.

[0020] The interruption control circuit 120 is a circuit for opening and closing the contact 110, and is configured by connecting an operation switch 122, a trip coil 123, and a palette switch 124 in series between the positive and negative electrodes of a DC power source 121.

[0021] The operating switch 122 is a switch that closes when the trip coil 123 is excited.

[0022] The trip coil 123 is a coil that is excited when the contact 110 is opened. In FIG.

[0023] The pallet switch 124 is a switch that is linked to the opening and closing of the contact 110. Specifically, the pallet switch 124 is closed when the contact 110 is closed, and opens when the contact 110 is opened to stop excitation of the trip coil 123.

[0024] When a fault current is detected to flow in the transmission / distribution line and the operation switch 122 is closed, the trip control circuit 120 becomes a closed circuit, a control current flows to the trip coil 123, and the trip coil 123 is excited. When the mechanism for opening and closing the contact 110 (such as the plunger 130 and latch 140 described below) operates in response to the excitation of the trip coil 123, the contact 110 opens from a closed state and the fault current is interrupted. Thereafter, the pallet switch 124 opens, stopping the excitation of the trip coil 123, and the contact 110 closes and closes again. At this time, the operation switch 122 opens again, and the contact 110 and the trip control circuit 120 go into a standby state for interrupting the fault current.

[0025] FIG. 2 is a graph showing a change in the control current flowing through the trip coil 123 while the trip coil 123 is excited. In the graph of FIG. 2, the horizontal axis indicates time (mS), and the vertical axis indicates the magnitude of the control current (A). The period from the start of excitation of the trip coil 123 to the end of excitation is divided into four periods (1) to (4) according to the value of the control current. The control current increases from 0 A during periods (1) to (3) and decreases to 0 A during period (4). In addition, below the horizontal axis of the graph, the relative positional relationship between the trip coil 123 and the plunger 130 and latch 140, which are mechanisms for opening and closing the contact 110, during each of periods (1) to (4) is shown. The plunger 130 is a member disposed inside the trip coil 123, which is displaced upward on the paper when the trip coil 123 is excited, and displaced downward on the paper when the excitation of the trip coil 123 is stopped. The latch 140 is a member that engages with the plunger 130. When the plunger 130 is displaced upward on the paper, the latch 140 is pushed upward on the paper, and when the plunger 130 is displaced downward on the paper, the latch 140 is released from engagement with the plunger 130 and returned to the position before being pushed upward on the paper.

[0026] Period (1) is the period during which operation switch 122 is closed and excitation of trip coil 123 begins. During this period (1), plunger 130 is not displaced upward in the drawing, and therefore latch 140 is not yet pushed upward in the drawing.

[0027] Period (2) is a period following period (1) and is a period during which excitation of trip coil 123 continues. During period (2), plunger 130 starts to be displaced upward in the drawing, and latch 140 is pushed upward in the drawing.

[0028] Period (3) is a period following period (2) and is a period during which the excitation of trip coil 123 continues. During period (3), plunger 130 is further displaced upward in the drawing, and latch 140 is further pushed upward in the drawing. Then, direct or indirect engagement between latch 140 and contact 110 is released, opening contact 110 and interrupting the fault current flowing through the power transmission / distribution line.

[0029] Period (4) is a period following period (3) during which excitation of trip coil 123 is stopped. During period (4), pallet switch 124 opens in conjunction with the opening of contact 110 during period (3), and excitation of trip coil 123 is stopped. This causes plunger 130 to be displaced downward on the page, and latch 140 is returned to the position it was in before it was pushed up. Then, latch 140 and contact 110 directly or indirectly engage with each other, closing contact 110, and the power transmission / distribution line returns to a state in which current flows.

[0030] Each of periods (1) to (4) may be extended due to slower operation of plunger 130 and latch 140 as a result of deterioration over time of the grease applied to the mechanical parts of circuit breaker 100. The degree of this extension varies depending on the degree of deterioration of the grease over time. Furthermore, period (4) may be extended due to an abnormality in the contacts of pallet switch 124 (e.g., roughness, oxide adhesion, etc.). The degree of this extension varies depending on the degree of abnormality in the contacts of pallet switch 124.

[0031] When the period (1) is extended, for example, deterioration of grease for smoothing the operation of the plunger 130 is considered to be a cause of the extension. When the period (2) is extended, for example, deterioration of grease for smoothing the operation of at least one of the plunger 130 and the latch 140 is considered to be a cause of the extension. When the period (3) is extended, for example, deterioration of grease for smoothing the operation of at least one of the plunger 130 and the latch 140, deterioration of grease for smoothly transmitting the operation of the latch 140 to the contact 110, etc. are considered to be a cause of the extension. When the period (4) is extended, for example, deterioration of grease for smoothing the operation of at least one of the plunger 130 and the latch 140, deterioration of grease for smoothly transmitting the operation of the latch 140 to the contact 110, contact abnormality of the pallet switch 124, etc. are considered to be a cause of the extension. In this way, the causes of the extension of each of the periods (1) to (4) are considered to be the causes of the extension.

[0032] 3 is a block diagram showing a schematic configuration of a prediction system 200 including an information processing device 300 according to this embodiment, which predicts the aging time when the circuit breaker 100 will malfunction and the cause of the malfunction of the circuit breaker 100 using a learning model 343. Note that a malfunction in the circuit breaker 100 refers to a state in which at least one of the periods (1) to (4) of the control current flowing through the trip coil 123 is extended due to the influence of aging deterioration of the grease in the circuit breaker 100, for example, and the circuit breaker 100 cannot operate normally, and includes a state in which the circuit breaker 100 does not operate.

[0033] For example, when an accident such as a short circuit or a ground fault occurs in a power transmission / distribution line, a fault current due to this accident will flow in the power transmission / distribution line. In order to interrupt this fault current, the circuit breaker 100 needs to open the closed contact 110 only for the period (1) to (4) in FIG. 2. In order to open the contact 110, a control current shown in the period (1) to (4) in FIG. 2 flows in the trip coil 123. The prediction system 200 is a system that, for example, acquires a waveform (third waveform) of the control current flowing in the trip coil 123 when the circuit breaker 100 currently interrupts the fault current, and predicts the aging time when the circuit breaker 100 will malfunction and the cause of the circuit breaker 100 malfunctioning based on this waveform using AI (Artificial Intelligence). The prediction system 200 is a system that outputs to relevant parties the time over which the circuit breaker 100 will malfunction and the causes of the circuit breaker 100 malfunctioning, thereby enabling the circuit breaker 100 to interrupt the fault current if an accident such as a short circuit or a ground fault occurs in the power transmission / distribution line.

[0034] The prediction system 200 includes an information processing device 300 and an external computer 400 as means for realizing the above functions. The information processing device 300 and the external computer 400 are connected in a state in which two-way communication is possible via a communication network 500. The communication network 500 is, for example, a local area network (LAN), a wide area network (WAN), a dedicated line, a power line communication network, various public communication networks, etc.

[0035] The information processing device 300 includes an input unit 310 , a prediction unit 320 , an output unit 330 , and a storage unit 340 .

[0036] The memory unit 340 stores a control program 341, learning data 342, and a learning model 343.

[0037] The control program 341 is a program executed by the information processing device 300 when making the prediction unit 320 predict the aging time when the circuit breaker 100 will currently malfunction and the cause of the circuit breaker 100 currently malfunctioning during that aging time based on the waveform (third waveform) of the control current currently flowing through the trip coil 123 of the circuit breaker 100 in the case where, for example, an accident occurs in the power transmission / distribution line, causing a fault current to flow in the power transmission / distribution line, and the circuit breaker 100 currently opens the contacts 110 to interrupt the fault current. Note that, in this embodiment, the timing of acquiring the waveform of the control current is when an accident occurs in the power transmission / distribution line, but it may be when no accident occurs in the power transmission / distribution line (for example, when inspecting the operation of the circuit breaker 100).

[0038] The learning data 342 is supervised data in which feature amounts and labels, which are past results, are associated with each other, and which is prepared in advance to allow the prediction unit 320 to predict the aging period at which the circuit breaker 100 will currently malfunction and the factors that will cause the circuit breaker 100 to currently malfunction at that aging period. In this embodiment, the learning data 342 includes first learning data 342A in which a first feature amount and a first label are associated with each other, and second learning data 342B in which a second feature amount and a second label are associated with each other, but may include only the first learning data 342A.

[0039] The first learning data 342A is learning data that is prepared in advance for, for example, a plurality of circuit breakers 100 that have the same specifications as the current circuit breaker 100 and were manufactured in the same manufacturing year, so that the learning model 343 can predict the time over time at which the current circuit breaker 100 will malfunction and the factors that will cause the circuit breaker 100 to malfunction at that time.

[0040] Fig. 4A is a diagram showing an example of waveform data included in the first feature of the first learning data 342A. Each of the waveform data D1 to D9 shown in Fig. 4A is data including, in the total aging period of 10 years, any of the following: (A) waveform data of the actually measured control current; (B) waveform data predicted based on the waveform of the actually measured control current of A; (C) data showing a malfunction of the circuit breaker 100 when the control current is not actually measured for the first time; and (D) data showing that a malfunction of the circuit breaker 100 is predicted based on the data of C. Furthermore, the frames in which the waveform data D1 to D9 correspond to each aging period of 1 year to 10 years are divided into any of a thick solid frame, a thin solid frame, a double frame of a thin solid line, and a double frame of a thin solid line and a thin dashed line. The thick solid line frame shows (A) waveform data of the actually measured control current, the thin solid line frame shows (B) waveform data of the predicted control current, the thin solid line double frame shows (C) data indicating a malfunction of the circuit breaker 100, and the thin solid line and thin dashed line double frame shows (D) data indicating that a malfunction of the circuit breaker 100 has been predicted.

[0041] The first characteristic amount is waveform data D1 to D9 including waveforms (first waveform, second waveform) of control currents flowing in the past to the trip coils 123 of the plurality of circuit breakers 100 during periods (1) to (4) in FIG. 2. The first waveform is a waveform of a control current acquired by intentionally closing the operation switch 122 regardless of the presence or absence of an accident on the power transmission / distribution line every first number of years (e.g., one year) from the year when each of the plurality of circuit breakers 100 was installed at the interruption position of the fault current. The second waveform is a waveform of a control current acquired by intentionally closing the operation switch 122 regardless of the presence or absence of an accident on the power transmission / distribution line every second number of years (e.g., two years, three years, etc.) different from the first number of years from the year when each of the plurality of circuit breakers 100 was installed at the interruption position of the fault current. Note that since it is only necessary to know the number of years since the installation of the plurality of circuit breakers 100, the plurality of circuit breakers 100 may be installed in the same year or in different years in an unused state. A plurality of circuit breakers 100 are assigned to each of the waveform data D1 to D9. Specifically, when, for example, 900 circuit breakers 100 are installed, 100 circuit breakers 100 are assigned to each of the waveform data D1 to D9 in order to acquire the waveform data D1 to D9. Each of the waveform data D1 to D9 includes nine pieces of data for each number of years since the installation of the circuit breakers 100, and data indicating the number of years since the installation of the circuit breakers 100 is associated with these nine pieces of data. Data indicating the lengths of the periods (1) to (4) are associated with the waveform data of the actually measured control current and the waveform data of the predicted control current included in the nine pieces of data for each number of years in each of the waveform data D1 to D9.

[0042] The waveform data D1 is data acquired with the tripping control circuit 120 as a closed circuit every time one year passes since the year of installation of each of the multiple circuit breakers 100. Of the waveform data D1, the data within the thick solid line frames corresponding to each of the aging years of 1 to 10 years is assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100 for ease of explanation.

[0043] The waveform data D2 is data acquired with the tripping control circuit 120 as a closed circuit every two years since the year of installation of each of the multiple circuit breakers 100. Of the waveform data D2, the data within the thick solid line frames corresponding to aging of 1 year, 3 years, 5 years, 7 years, and 9 years is assumed, for ease of explanation, to be waveform data of the control current actually measured from all of the multiple circuit breakers 100.

[0044] The waveform data D3 is data acquired with the tripping control circuit 120 as a closed circuit every time three years have passed since the year in which each of the multiple circuit breakers 100 was installed. For ease of explanation, the data within the thick solid line frames corresponding to the ages of 1 year, 4 years, 7 years, and 10 years among the waveform data D3 is assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100.

[0045] The waveform data D4 is data acquired with the tripping control circuit 120 as a closed circuit every time four years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, the data in the thick solid line frames corresponding to the aging years of one year and five years, respectively, among the waveform data D4, are assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100, and the data in the thin solid line double frame when the aging year is nine years is assumed to be data indicating that no control current was actually measured from all of the multiple circuit breakers 100. In other words, the waveform data D4 indicates that the multiple circuit breakers 100 that are at least nine years old or older will definitely malfunction.

[0046] The waveform data D5 is data acquired with the tripping control circuit 120 as a closed circuit every five years from the year of installation of each of the multiple circuit breakers 100. For ease of explanation, the data within the thick solid line frames corresponding to the years of 1 year and 6 years, respectively, of the waveform data D5, is assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100.

[0047] The waveform data D6 is data acquired with the tripping control circuit 120 as a closed circuit every six years from the year of installation of each of the multiple circuit breakers 100. Of the waveform data D6, the data within the thick solid line frames corresponding to the years of 1 year and 7 years, respectively, are assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100 for ease of explanation.

[0048] The waveform data D7 is data acquired with the tripping control circuit 120 as a closed circuit every time seven years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, among the waveform data D7, data within a thick solid line frame when the age is one year is waveform data of control currents actually measured from all of the multiple circuit breakers 100, and data within a thin solid double frame when the age is eight years is data indicating that no control current was actually measured from any of the multiple circuit breakers 100. In other words, the waveform data D7 indicates that the multiple circuit breakers 100 that are at least eight years old will definitely malfunction.

[0049] The waveform data D8 is data acquired with the tripping control circuit 120 as a closed circuit every time eight years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, among the waveform data D8, data within a thick solid line frame when the age is one year is waveform data of control currents actually measured from all of the multiple circuit breakers 100, and data within a thin solid double frame when the age is nine years is data indicating that no control currents were actually measured from all of the multiple circuit breakers 100. In other words, the waveform data D8 indicates that multiple circuit breakers 100 that are at least nine years old will definitely malfunction.

[0050] The waveform data D9 is data acquired with the tripping control circuit 120 as a closed circuit every time nine years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, among the waveform data D9, data within a thick solid line frame when the age is one year is waveform data of control currents actually measured from all of the multiple circuit breakers 100, and data within a thin solid double frame when the age is 10 years is data indicating that no control current was actually measured from all of the multiple circuit breakers 100. In other words, the waveform data D9 indicates that the multiple circuit breakers 100 that are at least 10 years old or older will definitely malfunction.

[0051] For example, the data periods (1) to (4) for the waveform data D1, D2, and D4 when the aging age is 5 years will be compared. For convenience of explanation, it is assumed that the circuit breaker 100 has not been operated for the purpose of interrupting a fault current after installation of the circuit breaker 100. The data for the waveform data D1 when the aging age is 5 years is acquired in the fifth operation of the multiple circuit breakers 100 assigned to acquire the waveform data D1, the data for the waveform data D2 when the aging age is 5 years is acquired in the third operation of the multiple circuit breakers 100 assigned to acquire the waveform data D2, and the data for the waveform data D4 when the aging age is 5 years is acquired in the second operation of the multiple circuit breakers 100 assigned to acquire the waveform data D4. The grease applied to the mechanical parts of the circuit breaker 100 deteriorates more as the number of times the circuit breaker 100 is operated increases, so the periods (1) to (4) of the data when the aging period is 5 years in the waveform data D1, D2, and D4 are longer in the order of the waveform data D1, D2, and D4. In this way, even if the data has the same aging period, the period (1) to (4) is longer as the number of times the circuit breaker 100 is operated since the installation of the circuit breaker 100 decreases. Also, in each of the waveform data D1 to D9, the periods (1) to (4) of the data for each aging period are longer as the aging period increases. In consideration of these characteristics, the waveform data in the thin solid line frames in each of the waveform data D1 to D9 are predicted using, for example, artificial intelligence, based on the waveform data in the thick solid line frames in the waveform data D1 to D9. In addition, the data predicted to be abnormal within the double frame of thin solid line and thin dashed line in each of the waveform data D5, D7, D8, and D9 is predicted using, for example, artificial intelligence, based on the waveform data within the thick solid line frame and thin solid line frame in the waveform data D1 to D9 and the data indicating the abnormality within the double frame of thin solid line in the waveform data D4, D7, D8, and D9.

[0052] The first label is data indicating a time (e.g., a first aging period) within the aging period of waveform data D1 when the waveform of the control current that should flow through trip coil 123 becomes unidentifiable for the first time, or when it is predicted that the waveform of the control current will become unidentifiable for the first time, even though operation switch 122 of shutoff control circuit 120 is closed to make shutoff control circuit 120 a closed circuit, and a time (e.g., a second aging period) within the aging periods of waveform data D2 to D9 when the waveform of the control current becomes unidentifiable for the first time, or when it is predicted that the waveform of the control current will become unidentifiable for the first time. In Fig. 4A, the data in the waveform data D4 indicating that the circuit breaker 100 having an aging age of 9 years is malfunctioning, the data in the waveform data D5 indicating that the circuit breaker 100 having an aging age of 10 years is predicted to be malfunctioning, the data in the waveform data D7 indicating that the circuit breaker 100 having an aging age of 8 years is malfunctioning, the data in the waveform data D8 indicating that the circuit breaker 100 having an aging age of 8 years is predicted to be malfunctioning, and the data in the waveform data D9 indicating that the circuit breaker 100 having an aging age of 8 years is predicted to be malfunctioning correspond to the first label. Note that in Fig. 4A, there is no data indicating that the circuit breaker 100 is malfunctioning or data indicating that the circuit breaker 100 is predicted to be malfunctioning in the waveform data D1, D2, D3, and D6. In such a case, when the aging period in the waveform data D1, D2, D3, and D6 is 11 years or more, based on the waveform data within the thick solid line frames and thin solid line frames in the waveform data D1 to D9, the data indicating that the circuit breaker 100 is malfunctioning within the double thin solid line frame in the waveform data D4, D7, D8, and D9, and the data indicating that the circuit breaker 100 is predicted to be malfunctioning within the double thin solid line and thin dashed line frame in the waveform data D5, D7, D8, and D9, the aging period when the circuit breaker 100 will first malfunction can be predicted, for example using artificial intelligence, and the data indicating this predicted aging period can be used as the first label.

[0053] Among the waveform data D1 to D9, the waveform data within the thick solid line frame and the thin solid line frame contain information indicating the lengths of the periods (1) to (4). Therefore, when the waveform data of the control current currently flowing through the trip coil 123 of the circuit breaker 100 is elongated, the cause of the elongation can be predicted based on the first learning data 342A.

[0054] The second learning data 342B is learning data prepared in advance for a plurality of circuit breakers 100 installed nationwide, for example, divided into regions such as by region (Hokkaido, Tohoku, Kanto, Chubu, Kinki, Chugoku, Shikoku, Kyushu, Okinawa) or by prefecture, so that the learning model 343 can predict the aging period when the current circuit breaker 100 will malfunction and the factors that will cause the current circuit breaker 100 to malfunction at that aging period. In this embodiment, the second learning data 342B is eight pieces of data prepared for each region. In addition, the plurality of circuit breakers 100 are, for example, the same type of circuit breaker as the current circuit breaker 100 (any of a vacuum circuit breaker, an oil circuit breaker, a porcelain circuit breaker, an air circuit breaker, and a gas circuit breaker).

[0055] Fig. 4B is a diagram showing an example of the second feature amount in the second learning data 342B, and is assumed to be data on a plurality of circuit breakers 100 installed in the Kanto region, for example, as one of eight pieces of data prepared for each region. Each of the waveform data D11 to D19 shown in Fig. 4B is data including, in the total age of 10 years, any of the following: (A) waveform data of the actually measured control current; (B) waveform data predicted based on the waveform of the actually measured control current of A; (C) data indicating that the control current is not actually measured for the first time and that the circuit breaker 100 is malfunctioning; and (D) data indicating that the circuit breaker 100 is predicted to be malfunctioning based on the data of C. Furthermore, the frames in which the waveform data D11 to D19 are associated with each age of 1 year to 10 years are divided into any of a thick solid frame, a thin solid frame, a double frame of a thin solid line, and a double frame of a thin solid line and a thin dashed line. The thick solid line frame shows (A) waveform data of the actually measured control current, the thin solid line frame shows (B) waveform data of the predicted control current, the thin solid line double frame shows (C) data indicating that the circuit breaker 100 is malfunctioning, and the thin solid line and thin dashed line double frame shows (D) data indicating that the circuit breaker 100 is predicted to be malfunctioning.

[0056] The second characteristic amount is waveform data D11 to D19 including waveforms (first waveform, second waveform) of control currents flowing in the trip coils 123 of the plurality of circuit breakers 100 in the past during periods (1) to (4) in FIG. 2. The first waveform is a waveform of a control current acquired by intentionally closing the operation switch 122 regardless of the presence or absence of an accident on the power transmission / distribution line every first number of years (e.g., one year) from the year when each of the plurality of circuit breakers 100 was installed at the interruption position of the fault current. The second waveform is a waveform of a control current acquired by intentionally closing the operation switch 122 regardless of the presence or absence of an accident on the power transmission / distribution line every second number of years (e.g., two years, three years, etc.) different from the first number of years from the year when each of the plurality of circuit breakers 100 was installed at the interruption position of the fault current. Note that since it is only necessary to know the number of years since the installation of the plurality of circuit breakers 100, the plurality of circuit breakers 100 may be installed in the same year or in different years in an unused state. A plurality of circuit breakers 100 are assigned to each of the waveform data D11 to D19. Specifically, in a case where, for example, 900 circuit breakers 100 are installed in the Kanto region, 100 circuit breakers 100 are assigned to each of the waveform data D11 to D19 in order to acquire the waveform data D11 to D19. Each of the waveform data D11 to D19 includes nine pieces of data for each number of years since the installation of the circuit breakers 100, and data indicating the number of years since the installation of the circuit breakers 100 is associated with these nine pieces of data. Furthermore, data indicating the lengths of the periods (1) to (4) are associated with the waveform data of the actually measured control current and the waveform data of the predicted control current included in the nine pieces of data for each number of years in each of the waveform data D11 to D19. Moreover, data representing the installation positions of the plurality of circuit breakers 100, such as data representing latitude and longitude, is associated with each of the waveform data D11 to D19.

[0057] The waveform data D11 is data acquired with the tripping control circuit 120 as a closed circuit every time one year has passed since the year of installation of each of the multiple circuit breakers 100. For convenience of explanation, the data in the thick solid line frames corresponding to the aging years of 1 to 9 years among the waveform data D11 is waveform data of the control current actually measured from all of the multiple circuit breakers 100, and the data in the thin solid line double frame when the aging year is 10 years is data indicating that no data was actually measured from all of the multiple circuit breakers 100. In other words, the waveform data D11 indicates that the multiple circuit breakers 100 that are at least 10 years old or older will definitely malfunction.

[0058] The waveform data D12 is data acquired with the tripping control circuit 120 as a closed circuit every two years since the year of installation of each of the multiple circuit breakers 100. Of the waveform data D12, the data within the thick solid line frames corresponding to aging of 1 year, 3 years, 5 years, 7 years, and 9 years is assumed, for ease of explanation, to be waveform data of the control current actually measured from all of the multiple circuit breakers 100.

[0059] The waveform data D13 is data acquired with the tripping control circuit 120 as a closed circuit every time three years have passed since the year of installation of each of the multiple circuit breakers 100. For convenience of explanation, the data within the thick solid line frames corresponding to the aging years of one year, four years, and seven years, respectively, of the waveform data D13, is waveform data of the control current actually measured from all of the multiple circuit breakers 100, and the data within the thin solid line double frame when the aging year is ten years is data indicating that no control current was actually measured from all of the multiple circuit breakers 100. In other words, the waveform data D13 indicates that the multiple circuit breakers 100 that are at least aged for ten years or more will definitely malfunction.

[0060] The waveform data D14 is data acquired with the tripping control circuit 120 as a closed circuit every time four years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, the data in the thick solid line frames of the waveform data D4 corresponding to the years of age of one year and five years are waveform data of the control current actually measured from all of the multiple circuit breakers 100, and the data in the thin solid line double frame when the year of age is nine years is data indicating that no control current was actually measured from any of the multiple circuit breakers 100. In other words, the waveform data D14 indicates that the multiple circuit breakers 100 that are at least nine years old will definitely malfunction.

[0061] The waveform data D15 is data acquired with the tripping control circuit 120 as a closed circuit every five years from the year of installation of each of the multiple circuit breakers 100. Of the waveform data D15, the data within the thick solid line frames corresponding to the years of one and six, respectively, is assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100 for ease of explanation.

[0062] The waveform data D16 is data acquired with the tripping control circuit 120 as a closed circuit every six years from the year of installation of each of the multiple circuit breakers 100. Of the waveform data D16, the data within the thick solid line frames corresponding to the years of one and seven, respectively, is assumed to be waveform data of the control current actually measured from all of the multiple circuit breakers 100 for ease of explanation.

[0063] The waveform data D17 is data acquired with the tripping control circuit 120 as a closed circuit every time seven years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, among the waveform data D17, data within a thick solid line frame when the age is one year is waveform data of control currents actually measured from all of the multiple circuit breakers 100, and data within a thin solid double frame when the age is eight years is data indicating that no control current was actually measured from any of the multiple circuit breakers 100. In other words, the waveform data D17 indicates that the multiple circuit breakers 100 that are at least eight years old will definitely malfunction.

[0064] The waveform data D18 is data acquired with the tripping control circuit 120 as a closed circuit every time eight years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, among the waveform data D18, data within a thick solid line frame when the age is one year is waveform data of control currents actually measured from all of the multiple circuit breakers 100, and data within a thin solid double frame when the age is nine years is data indicating that no control current was actually measured from any of the multiple circuit breakers 100. In other words, the waveform data D18 indicates that multiple circuit breakers 100 that are at least nine years old will definitely malfunction.

[0065] The waveform data D19 is data acquired with the tripping control circuit 120 as a closed circuit every time nine years have passed since the year of installation of each of the multiple circuit breakers 100. For ease of explanation, among the waveform data D19, data within a thick solid line frame when the age is one year is waveform data of control currents actually measured from all of the multiple circuit breakers 100, and data within a thin solid double frame when the age is 10 years is data indicating that no control currents were actually measured from all of the multiple circuit breakers 100. In other words, the waveform data D19 indicates that the multiple circuit breakers 100 that are at least 10 years old or older will definitely malfunction.

[0066] For the waveform data D11-19, when the number of years of aging is the same, the fewer the number of operations of the circuit breaker 100 since its installation, the longer the lengths of the periods (1)-(4) in Fig. 2 become. Also, in each of the waveform data D11-D19, the longer the number of years of aging, the longer the periods (1)-(4) of the data for each number of years of aging become. In light of these characteristics, the waveform data within the thin solid line frames in each of the waveform data D11-D19 is predicted, for example, using artificial intelligence, based on the waveform data within the thick solid line frames in the waveform data D11-D19. In addition, the data predicted to be abnormal within the double frame of thin solid line and thin dashed line in each of the waveform data D12, D14 to D19 is predicted using, for example, artificial intelligence based on the waveform data within the thick solid line frame and thin solid line frame in the waveform data D11 to D19 and the data indicating abnormality within the double frame of thin solid line in the waveform data D10, D13, D14, D17 to D19.

[0067] The second label is data indicating a time period (e.g., a first aging period) within the aging period of the waveform data D11 when the waveform of the control current that should flow through trip coil 123 becomes unidentifiable for the first time, or when it is predicted that the waveform of the control current will become unidentifiable for the first time, even though the operating switch 122 of the shutoff control circuit 120 is closed to make the shutoff control circuit 120 a closed circuit, and a time period (e.g., a second aging period) within the aging periods of the waveform data D12 to D19 when the waveform of the control current becomes unidentifiable for the first time, or when it is predicted that the waveform of the control current will become unidentifiable for the first time. In FIG. 4B , there are waveform data D11 showing data indicating that the circuit breaker 100 having an aging age of 10 years is malfunctioning, waveform data D12 showing data indicating that the circuit breaker 100 having an aging age of 10 years is predicted to be malfunctioning, waveform data D13 showing data indicating that the circuit breaker 100 having an aging age of 10 years is malfunctioning, waveform data D14 showing data indicating that the circuit breaker 100 having an aging age of 9 years is malfunctioning, and waveform data D15 showing data indicating that the circuit breaker 100 having an aging age of 8 years is malfunctioning. The second label includes data indicating that the circuit breaker 100 is predicted to malfunction when the aging age is 9 years in the waveform data D16, data indicating that the circuit breaker 100 is predicted to malfunction when the aging age is 8 years in the waveform data D17, data indicating that the circuit breaker 100 is predicted to malfunction when the aging age is 8 years in the waveform data D18, and data indicating that the circuit breaker 100 is predicted to malfunction when the aging age is 8 years in the waveform data D19.

[0068] Among the waveform data D11 to D19, the waveform data within the thick solid line frame and the thin solid line frame contain information indicating the lengths of the periods (1) to (4). Therefore, when the waveform data of the control current currently flowing through the trip coil 123 of the circuit breaker 100 is elongated, the cause of the elongation can be predicted based on the second learning data 342B.

[0069] The learning model 343 is a machine learning model formed to predict the aging time when the circuit breaker 100 will malfunction and the cause of the malfunction at that aging time when waveform data of the control current flowing through the trip coil 123 of the circuit breaker 100 at present is input by learning using the first learning data 342A and the second learning data 342B. The more the learning model 343 learns using the first learning data 342A and the second learning data 342B, the more accurate the prediction becomes. In this embodiment, the learning model 343 is, for example, a deep neural network (DNN), but may be another type of model such as a gradient boosting decision tree (GBDT).

[0070] FIG. 5 is a diagram showing the structure of a deep neural network, which is an example of the learning model 343.

[0071] The learning model 343 includes three layers: an input layer 3431, an intermediate layer 3432, and an output layer 3433.

[0072] The input layer 3431 is a layer to which waveform data, which is the same type as the first feature amount constituting the first learning data 342A and the second feature amount constituting the second learning data 342B, and indicates the waveform of the control current flowing through the trip coil 123 of the circuit breaker 100 at present, is input. The intermediate layer 3432 is a layer including one or more hidden layers consisting of one or more nodes including parameters adjusted by performing learning using the first learning data 342A and the second learning data 342B. The intermediate layer 3432 predicts the aging time when the circuit breaker 100 will malfunction at present and the cause of the malfunction, based on the data input to the input layer 3431. The output layer 3433 is a layer that outputs the prediction result by the intermediate layer 3432.

[0073] For example, consider a case where a control current flows through the trip coil 123 of the current circuit breaker 100, which has been installed for five years, and waveform data obtained by actually measuring this control current is input to the input layer 3431 of the learning model 343. It is assumed that a control current has flowed through the trip coil 123 of the current circuit breaker 100 only once when the circuit breaker 100 was installed for one year, and that a control current has now flowed for the second time. It is also assumed that the current circuit breaker 100 is installed at a specified location in Tokyo.

[0074] In the intermediate layer 3432, based on the learning results using the first learning data 342A and the second learning data 342B, a prediction is made of the current aging period at which the circuit breaker 100 is most likely to malfunction and the factor that is most likely to cause the malfunction at that aging period. Specifically, the intermediate layer 3432 refers to the first learning data 342A and the second learning data 342B, and selects waveform data (any of D1 to D9, D11 to D19) to which a circuit breaker 100 having the same or similar history as the current circuit breaker 100 is assigned. Further, from the selected waveform data, data that is the same as or similar to the period (1) to (4) of the waveform data of the control current flowing through the trip coil 123 of the circuit breaker 100 at present (any of 1 to 10 years in age) is selected, and a time period over which the current circuit breaker 100 will malfunction is predicted based on the relationship with the first label and the second label, and the cause of the malfunction at that time period is predicted based on the difference with the period (1) to (4) of the waveform data of the control current flowing through the circuit breaker 100 at present.

[0075] For example, in the intermediate layer 3432, when data of the waveform data D4 when the aging age is 5 years is selected from the first learning data 342A, and data of the waveform data D14 when the aging age is 5 years is selected from the second learning data 342B, the aging time when the circuit breaker 100 will malfunction at present is predicted to be when the aging age is 9 years, based on a comprehensive judgment. Also, when any of the periods (1) to (4) of the waveform data of the control current flowing through the circuit breaker 100 at present is extended with respect to the periods (1) to (4) of the predicted data when the aging age is 8 years in the waveform data D4 and D14, the cause of the circuit breaker 100 malfunctioning is predicted according to the extended period. When the prediction result of the intermediate layer 3432 is output as a message, for example, a message such as "There is a high possibility that the circuit breaker will malfunction due to solidification of the grease when the aging age reaches 9 years. Please perform maintenance before the aging age reaches 8 years" is output.

[0076] Furthermore, for example, as shown in waveform data D7 to D7 in the first learning data 342A and waveform data D17 to D19 in the second learning data 342B, if the current age of the circuit breaker 100 is seven years but no control current flows through the trip coil 123, and the first learning data 342A and the second learning data 342B associate a label indicating a malfunction of the circuit breaker 100 with the following year, the intermediate layer 3432 may output a message such as "There is a high possibility that the circuit breaker will malfunction next year due to solidification of the grease. Please perform maintenance this year" when the circuit breaker 100 is seven years old.

[0077] 3, when the information processing device 300 predicts the current aging time when the circuit breaker 100 will malfunction and the cause of the circuit breaker 100 malfunctioning at that aging time using the learning model 343, the input unit 310 receives as input waveform data of the control current that has flowed through the trip coil 123 of the circuit breaker 100. At this time, data indicating the number of years since the installation of the circuit breaker 100, the number of times that the control current has flowed through the trip coil 123, and the installation position (latitude, longitude) of the circuit breaker 100 are also input to the input unit 310. The data input to the input unit 310 is then input to the prediction unit 320.

[0078] The prediction unit 320 inputs the above data input to the input unit 310 to the input layer 3431 of the learning model 343, and instructs the learning model 343 to make the above prediction.

[0079] The output unit 330 outputs the prediction result by the learning model 343 to the outside in accordance with an instruction from the prediction unit 320 triggered by the prediction result being output from the intermediate layer 3432 to the output layer 3433. Possible output methods of the output unit 230 include, for example, a method of displaying and outputting the prediction result on a display panel, a method of outputting the prediction result as sound from a speaker, a method of printing and outputting the prediction result on a paper medium, and the like.

[0080] The external computer 400 has a storage unit 410, stores data indicating prediction results by the learning model 343 in the storage unit 410, and reads out the data indicating the prediction results from the storage unit 410 as necessary and outputs it to the information processing device 300. Alternatively, a storage area other than the control program 341, the learning data 342, and the learning model 343 may be provided in the storage unit 340 of the information processing device 300, and the prediction results by the learning model 343 may be stored in this storage area.

[0081] FIG. 6 is a block diagram showing an example of hardware of an information processing device 300 used to realize the prediction system 200. As shown in FIG.

[0082] The information processing device 300 includes a processor 3010, a main memory device 3020, an auxiliary memory device 3030, an input device 3040, an output device 3050, and a communication device 3060. The information processing device 300 is, for example, a personal computer, an office computer, various server devices, a general-purpose machine, or the like. The information processing device 300 may be realized, in whole or in part, by using virtual information processing resources provided by using a virtualization technology, such as a virtual server provided by a cloud system. The prediction system 200 may be realized by using a plurality of information processing devices 300 connected to each other so as to be able to communicate with each other.

[0083] The processor 3010 is configured using, for example, a Central Processing Unit (CPU), a Micro Processing Unit (MPU), a Graphics Processing Unit (GPU), a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Artificial Intelligence (AI) chip, etc.

[0084] The main memory device 3020 is a device that stores programs and data, and is, for example, a Read Only Memory (ROM), a Random Access Memory (RAM), or a non-volatile memory (NVRAM (Non Volatile RAM)).

[0085] The auxiliary storage device 3030 is, for example, a solid state drive (SSD), a hard disk drive, an optical storage device (CD (Compact Disc), DVD (Digital Versatile Disc), etc.), a storage system, an IC card, an SD card, a read / write device for a recording medium such as an optical recording medium, a storage area of ​​a cloud server, etc. Programs and data can be read into the auxiliary storage device 3030 via a recording medium reader or a communication device 3060. The programs and data stored in the auxiliary storage device 3030 are read into the main storage device 3020 as needed.

[0086] The input device 3040 is an interface that accepts input from the outside, and is, for example, a keyboard, a mouse, a touch panel, a card reader, a pen-input tablet, a voice input device, or the like.

[0087] The output device 3050 is an interface that outputs various information such as the progress of processing and the results of processing. The output device 3050 is, for example, a display device (LCD (Liquid Crystal Display), graphic card, etc.) that visualizes the various information described above, a device that converts the various information described above into voice (voice output device (speaker, etc.)), or a device that converts the various information described above into text (printer, etc.). The information processing device 300 may be configured to input and output information to and from other devices via the communication device 3060.

[0088] The input device 3040 and the output device 3050 constitute a user interface that receives and presents information to and from the user.

[0089] The communication device 3060 is a device that realizes communication (wired communication or wireless communication) with other devices via a communication infrastructure such as a communication network 500, and is configured using, for example, a NIC (Network Interface Card), a wireless communication module, a USB module, etc.

[0090] The information processing device 300 may include, for example, an operating system, a file system, a data base management system (DBMS) (relational database, NoSQL, etc.), a key-value store (KVS), etc.

[0091] The functions of the prediction system 200 are realized by the processor 3010 of the information processing device 300 executing a program (control program) loaded into the main memory device 3020, or by the functions of the hardware (FPGA, ASIC, AI chip, etc.) constituting the prediction system 200. For example, the memory unit 340 in FIG. 3 is realized by the main memory device 3020 and the auxiliary memory device 3030, the input unit 310 in FIG. 3 is realized by the input device 3040, the prediction unit 320 in FIG. 3 is realized by the processor 3010, and the output unit 330 in FIG. 3 is realized by the output device 3050 and the communication device 3060.

[0092] FIG. 7 is a flowchart showing a process performed by the information processing device 300 to train the learning model 343 using the first learning data 342A and the second learning data 342B.

[0093] First, the information processing device 300 generates first learning data 342A in which the number of years since the control current first stopped flowing normally through the trip coil 123 of the plurality of circuit breakers 100 after the installation of the plurality of circuit breakers 100 or the number of years since the control current first stopped flowing normally through the trip coil 123 of the plurality of circuit breakers 100 after the installation of the plurality of circuit breakers 100 is associated as a first label to each of the waveform data D1 to D9 included in the first feature amount shown in Fig. 4A. Also, the information processing device 300 generates second learning data 342B in which the number of years since the control current first stopped flowing normally through the trip coil 123 of the plurality of circuit breakers 100 after the installation of the plurality of circuit breakers 100 or the number of years since the control current first stopped flowing normally through the trip coil 123 of the plurality of circuit breakers 100 after the installation of the plurality of circuit breakers 100 is associated as a second label to each of the waveform data D11 to D19 included in the second feature amount by region or prefecture shown in Fig. 4B (S1000).

[0094] Next, the information processing device 300 trains the learning model 343 using the first learning data 342A and the second learning data 342B (S1010). Note that the information processing device 300 may verify the prediction accuracy of the trained learning model 343. In this case, the information processing device 300 prepares learning data and verification data as the first learning data 342A and the second learning data 342B, respectively, trains the learning model 343 using the learning data, and verifies the learning model 343 using the verification data.

[0095] FIG. 8 is a flowchart showing a process in which the information processing device 300 uses the learning model 343 to predict the current aging time when the circuit breaker 100 will malfunction and the cause of the circuit breaker 100 malfunctioning at that aging time.

[0096] First, the information processing device 300 determines whether or not waveform data obtained by actually measuring the control current currently flowing through the trip coil 123 of the circuit breaker 100 has been input to the input unit 310 (S2000). When the information processing device 300 determines that the waveform data has not been input to the input unit 310 (S2000: NO), it executes the process of step S2000 again.

[0097] Next, when the information processing device 300 determines that the above waveform data has been input to the input unit 310 (S2000: YES), it uses the learning model 343 to predict the current aging period when the circuit breaker 100 will malfunction and the cause of the circuit breaker 100 malfunctioning at that aging period (S2010).

[0098] Next, the information processing device 300 outputs the prediction result by the learning model 343 from the output unit 330 (S2030).

[0099] As described above, the information processing device 300 includes a processor 3010 and a memory device (main memory device 3020, auxiliary memory device 3030) and predicts the aging time when the circuit breaker 100 including the trip coil 123 excited by the flow of a control current and the contacts 110 that interrupt the current flowing through the power transmission / distribution line when the trip coil 123 is excited will malfunction. The information processing device 300 predicts the aging time when the circuit breaker 100 including the trip coil 123 excited by the flow of a control current will malfunction by storing a first waveform, which is a waveform of a control current acquired every first aging period (for example, one year) in the past, and a second aging period different from the first aging period. A learning model 343 is stored, which performs learning based on learning data 342 that associates features indicating a second waveform, which is the waveform of the control current acquired every (e.g., every 2 years, 3 years, etc.), with a label indicating a first aging period when the first waveform cannot be acquired, and a label indicating a second aging period when the second waveform cannot be acquired.Currently, a third waveform, which is the waveform of the control current, is input to the learning model 343, and the third aging period when the third waveform cannot be acquired is predicted as the aging period when the circuit breaker 100 will malfunction.

[0100] According to the information processing device 300, maintenance of the circuit breaker 100 can be performed before the time when the circuit breaker 300 is predicted to malfunction, thereby enabling the circuit breaker 100 to interrupt a fault current flowing in a power transmission / distribution line.

[0101] Moreover, the information processing device 300 currently inputs the third waveform into the learning model 343 to predict the cause of the circuit breaker 100 malfunctioning as well as the time over time when the circuit breaker 100 will malfunction.

[0102] According to the information processing device 300, it is possible to perform maintenance focusing on factors that are predicted to cause the circuit breaker 100 to malfunction before the time when the circuit breaker 100 is predicted to malfunction, thereby enabling the circuit breaker to interrupt the fault current flowing in the power transmission and distribution line.

[0103] In addition, the circuit breakers 100 that are the subject of prediction in the information processing device 300 are multiple circuit breakers of the same specifications, the first waveform is the waveform of the control current in each of the multiple circuit breakers 100 of the same specifications, and the second waveform is the waveform of the control current in each of the multiple circuit breakers 100 of the same specifications.

[0104] According to the information processing device 300, the learning data is data in which a first aging period and a second aging period are associated with each of the first waveform and the second waveform obtained from multiple circuit breakers 100 of the same specifications. Therefore, by increasing the amount of learning data learned by the learning model 343, it becomes possible to accurately predict the aging period at which the circuit breaker 100 will malfunction.

[0105] In addition, the circuit breakers 100 that are the subject of prediction in the information processing device 300 are multiple circuit breakers 100 installed in multiple regions, the first waveform is the waveform of the control current in each of the multiple circuit breakers 100, and the second waveform is the waveform of the control current in each of the multiple circuit breakers 100.

[0106] According to the information processing device 300, the learning data includes information on the areas in which multiple circuit breakers 100 are installed, and thus the amount of learning data learned by the learning model 343 is further increased, making it possible to accurately predict the time over time when the circuit breakers 100 will malfunction.

[0107] The above-described embodiment is for the purpose of facilitating understanding of the present invention, and is not intended to limit the present invention. The present invention may be modified or improved without departing from the spirit of the present invention, and equivalents thereof are also included in the present invention. [Explanation of symbols]

[0108] 100 Circuit Breaker 110 Contacts 120 Shutdown control circuit 121 DC power supply 122 Operation switch 123 Trip Coil 124 Palette Switch 130 Plunger 140 Latch 200 Prediction System 300 Information processing device 310 Input section 320 Prediction Department 330 Output section 340,410 storage section 341 Control Program 342 Training Data 342A First training data 342B Second training data 343 Learning Model 400 External Computer 500 Communication Network 3010 Processor 3020 Main storage 3030 Auxiliary storage device 3040 Input Device 3050 output device 3060 Communication Equipment 3431 Input Layer 3432 Middle Class 3433 Output Layer

Claims

1. An information processing device having a processor and a storage device, and configured to predict an aging time when a circuit breaker including a trip coil that is excited when a control current flows and a contact that interrupts a current flowing through a power system when the trip coil is excited, comprising: storing a learning model that performs learning based on learning data in which feature quantities indicating a first waveform, which is a waveform of the control current acquired at each first aging period, and a second waveform, which is a waveform of the control current acquired at each second aging period different from the first aging period, are associated with labels indicating a first aging period when the first waveform could not be acquired and a second aging period when the second waveform could not be acquired; At present, a third waveform, which is the waveform of the control current, is input to the learning model, and a third aging time when the third waveform cannot be acquired is predicted as the aging time when the circuit breaker will malfunction. Information processing device.

2. 2. The information processing device according to claim 1, At present, the third waveform is input into the learning model to predict the time when the circuit breaker will malfunction and the cause of the circuit breaker malfunctioning. Information processing device.

3. 2. The information processing device according to claim 1, The circuit breaker is a plurality of circuit breakers of the same specification, the first waveform is a waveform of the control current in each of the plurality of circuit breakers having the same specifications, The second waveform is a waveform of the control current in each of the plurality of circuit breakers having the same specifications. Information processing device.

4. 4. The information processing device according to claim 3, The circuit breaker includes a plurality of circuit breakers installed in a plurality of regions, the first waveform is a waveform of the control current in each of the plurality of circuit breakers, The second waveform is a waveform of the control current in each of the plurality of circuit breakers. Information processing device.

5. An information processing device having a processor and a storage device, and configured to predict when a circuit breaker including a trip coil that is excited when a control current flows and a contact that interrupts a current flowing through a power system when the trip coil is excited, storing a learning model that performs learning based on learning data in which feature quantities indicating a first waveform, which is a waveform of the control current acquired at each first aging period, and a second waveform, which is a waveform of the control current acquired at each second aging period different from the first aging period, are associated with labels indicating a first aging period when the first waveform could not be acquired and a second aging period when the second waveform could not be acquired; At present, a third waveform, which is the waveform of the control current, is input to the learning model, and a third aging time when the third waveform cannot be acquired is predicted as the aging time when the circuit breaker will malfunction. Information processing methods.

6. 6. An information processing method according to claim 5, At present, the third waveform is input into the learning model to predict the time when the circuit breaker will malfunction and the cause of the circuit breaker malfunctioning. Information processing methods.

7. 6. An information processing method according to claim 5, The circuit breaker is a plurality of circuit breakers of the same specification, the first waveform is a waveform of the control current in each of the plurality of circuit breakers having the same specifications, The second waveform is a waveform of the control current in each of the plurality of circuit breakers having the same specifications. Information processing methods.

8. 8. An information processing method according to claim 7, The circuit breaker includes a plurality of circuit breakers installed in a plurality of regions, the first waveform is a waveform of the control current in each of the plurality of circuit breakers, The second waveform is a waveform of the control current in each of the plurality of circuit breakers. Information processing methods.

9. A computer having a processor and a storage device, which predicts an aging time when a circuit breaker including a trip coil that is excited when a control current flows and a contact that interrupts a current flowing through a power system when the trip coil is excited, a process of storing a learning model that learns based on learning data in which a feature quantity indicating a first waveform, which is a waveform of the control current acquired at each first aging period in the past, and a second waveform, which is a waveform of the control current acquired at each second aging period different from the first aging period, are associated with a label indicating a first aging period when the first waveform could not be acquired and a label indicating a second aging period when the second waveform could not be acquired; A process of predicting a third aging time when the third waveform, which is a waveform of the control current, cannot be acquired as the aging time when the circuit breaker will malfunction by inputting the third waveform, which is a waveform of the control current, into the learning model; A program that executes the following.

10. The program according to claim 9, At present, a process of predicting the time when the circuit breaker will malfunction and the cause of the circuit breaker malfunctioning by inputting the third waveform into the learning model. A program that executes the following.

11. The program according to claim 9, The circuit breaker is a plurality of circuit breakers of the same specification, the first waveform is a waveform of the control current in each of the plurality of circuit breakers having the same specifications, The second waveform is a waveform of the control current in each of the plurality of circuit breakers having the same specifications. program.

12. The program according to claim 11, The circuit breaker includes a plurality of circuit breakers installed in a plurality of regions, the first waveform is a waveform of the control current in each of the plurality of circuit breakers, The second waveform is a waveform of the control current in each of the plurality of circuit breakers. program.

Images