Machine learning device and opening / closing operation procedure determination device

Abstract

To provide a mechanical learning device capable of constructing a learning model for more efficiently determining a switching operation order of a section switch at the power failure work of a power system, and provide a switching operation order determination device using the learning model.SOLUTION: A mechanical learning device 1 comprises: an input data acquisition part 11 that acquires a data group containing a power failure object section X in a power system 3 that can adjust load currents of power distribution systems A to C, a time zone of a power failure operation for stopping a power in the power failure object section X, the weather at the power failure operation, and a loading current in each of a plurality of sections 7 in the time zone as input data; a label acquisition part 12 that acquires, as a label, a switch operation order of a section switch 6 for adjusting the loading current of each of the power distribution systems A to C at the power failure operation; and a learning model construction part 14 that constructs a learning model for determining the switch operation order of the section switch 6 at the power failure operation by performing a learning with a teacher by using the group of the input data and the labeling as teacher data.SELECTED DRAWING: Figure 7

Description

Technical Field

[0001] The present invention relates to a machine learning device and an opening / closing operation procedure determination device.

Background Art

[0002] Conventionally, in a power system including a plurality of distribution systems, a technique for adjusting the load current of each distribution system by operating sectional switches or the like is known. Patent Documents 1 and 2 describe this type of technique. Patent Document 1 describes a technique for obtaining the reserve capacity for each distribution system from the difference between the allowable value of each distribution system and the load current flowing through each distribution system, and opening and closing the sectional switch or the like based on the reserve capacity. Patent Document 2 describes a technique for changing the distribution line to an important load by loop switching of the distribution line when the occurrence section of abnormal weather is included in the distribution line to the important load.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, when performing a power outage operation on a power system, the supply of power to a section downstream of the power outage target section is hindered. Therefore, it is necessary to open and close the sectional switch so that power can be supplied from a different distribution system to the section. However, since the type of load and the magnitude of the load current also differ for each section, and the load current of each section also changes depending on conditions such as the time zone of the power outage operation, it takes time to determine an appropriate opening / closing operation procedure for the sectional switch for each power outage operation. Although the techniques of Patent Documents 1 and 2 can adjust the load current of each distribution system, there is room for improvement in terms of more efficiently determining the opening / closing operation procedure of the sectional switch.

[0005] An object of the present invention is to provide a machine learning device capable of constructing a learning model for more efficiently determining an opening / closing operation procedure of a sectional disconnector during a power outage operation of a power system, and an opening / closing operation procedure determination device using the learning model.

Means for Solving the Problems

[0006] The present invention relates to a machine learning device including: an input data acquisition unit that acquires, as input data, a data group including a power outage target section in a power system capable of adjusting load currents of respective power distribution systems by opening and closing sectional disconnectors that divide a distribution line into a plurality of sections, a time zone of a power outage operation for powering off the power outage target section, weather during the power outage operation, and load currents of the respective sections in the plurality of sections during the time zone; a label acquisition unit that acquires, as a label, an opening / closing operation procedure of the sectional disconnector for adjusting load currents of the respective power distribution systems during the power outage operation; and a learning model construction unit that constructs a learning model for determining the opening / closing operation procedure of the sectional disconnector during the power outage operation by performing supervised learning using a set of the input data and the label as supervised data.

[0007] The input data acquisition unit further acquires, as input data, power generation amounts by distributed power sources arranged in respective ones of the plurality of sections during the time zone.

[0008] The present invention also relates to an opening / closing operation procedure determination device using the learning model constructed by the machine learning device, including: a determination data acquisition unit that acquires determination data including a new power outage target section in the power system, a time zone of a new power outage operation for powering off the new power outage target section, weather during the new power outage operation, and load currents of the respective sections in the plurality of sections during the time zone of the new power outage operation; and an opening / closing operation procedure determination unit that determines an opening / closing operation procedure of the sectional disconnector during the new power outage operation based on the determination data and the learning model.

Effects of the Invention

[0009] According to the present invention, a machine learning device capable of constructing a learning model for more efficiently determining the opening / closing operation procedure of a sectional disconnector during a power system power outage operation, and an opening / closing operation procedure determination device using the learning model can be provided.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Embodiments for Carrying Out the Invention

[0011] Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing the overall configuration of an opening / closing operation procedure determination system 100. FIG. 2 is a schematic diagram showing an example of a power system 3 used in the opening / closing operation procedure determination system 100 according to the present embodiment.

[0012] The opening / closing operation procedure determination system 100 is a system for determining the opening / closing operation procedure of the sectionalizer 6 during a power outage operation in a power outage target section X of the power system 3. First, before explaining the opening / closing operation procedure determination system 100, an example of the power system 3 used by the opening / closing operation procedure determination system 100 will be described with reference to FIG. 2.

[0013] The power system 3 includes a first substation 5a, a second substation 5b, and a third substation 5c, which are three substations 5, a distribution line 4 provided between the three substations 5, three feeder circuit breakers 50, namely, feeder circuit breakers 51, 52, and 53, circuit breakers 81 and 82, an automatic voltage regulator (hereinafter referred to as SVR) 83, and a plurality of sectionalizers 6, namely, sectionalizers 60 to 67.

[0014] Further, the power system 3 is divided into three distribution systems A, B, and C. The distribution system A supplies power from the first substation 5a to the distribution line 4. The distribution system B supplies power from the second substation 5b to the distribution line 4. The distribution system C supplies power from the third substation 5c to the distribution line 4. Loads such as houses, stores, and factories that consume power, as well as distributed power sources 9 such as solar power generation, are connected to the distribution line 4.

[0015] The first substation 5a is provided with a feeder circuit breaker 51. The load current flowing through the distribution system A is limited by the feeder circuit breaker 51 to be equal to or less than a predetermined allowable value (hereinafter referred to as "the allowable load current of the distribution system A").

[0016] The second substation 5b is provided with a feeder circuit breaker 52. The load current flowing through the distribution system B is limited by the feeder circuit breaker 52 to be equal to or less than a predetermined allowable value (hereinafter referred to as "the allowable load current of the distribution system B").

[0017] The third substation 5c is provided with a feeder breaker 53. The load current flowing through the power distribution system C is limited by the feeder breaker 53 to a predetermined allowable value (hereinafter referred to as the "allowable load current of the power distribution system C") or less.

[0018] The distribution line 4 is divided into a plurality of sections 7 by circuit breakers 81, 82, SVR 83, and sectionalizing switches 6. For example, in the power system 3 shown in FIG. 2, a section 71 between the circuit breaker 81 and the sectionalizing switch 61, a section 71 between the sectionalizing switches 61, 62, 63, a section 73 between the sectionalizing switches 63, 64, 67, a section 74 between the sectionalizing switches 64, 65, a section 75 between the sectionalizing switches 65, 66, a section 76 between the sectionalizing switch 66 and the SVR 83, and a section 77 between the sectionalizing switch 67 and the circuit breaker 82. Although a plurality of sectionalizing switches 6 and sections 7 are provided between the second substation 5b and the SVR 83, they are not shown in FIGS. 2 to 6.

[0019] The circuit breaker 81 is a device that cuts off the power supply when the load current on its downstream side exceeds a predetermined allowable value (hereinafter referred to as the "allowable load current of the circuit breaker 81"). In this specification, the downstream side refers to the side opposite to each substation 5 in each power distribution system A to C.

[0020] The circuit breaker 82 is a device that cuts off the power supply when the load current on its downstream side exceeds a predetermined allowable value (hereinafter referred to as the "allowable load current of the circuit breaker 82").

[0021] The SVR 83 is a device that controls the voltage applied to the downstream distribution line 4. The SVR 83 is provided in the middle of the power distribution path to suppress voltage drop, and performs control to increase the voltage supplied from the substation 5 when the load current is large and decrease the voltage when the load current is small. An allowable value of the load current flowing through the downstream side of the SVR 83 (hereinafter referred to as the "allowable load current of the SVR 83") is set.

[0022] The power distribution systems A to C that supply power to each section 7 of the distribution line 4 are selected by opening and closing the sectional disconnectors 6. For example, in the state shown in Fig. 2, the sectional disconnectors 61 to 63 and 66 are closed, and the sectional disconnectors 65 and 67 are open. As a result, power is supplied from the power distribution system A to the sections 71 to 74, power is supplied from the power distribution system B to the sections 75 and 76, and power is supplied from the power distribution system C to the section 77. That is, the power system 3 is configured such that the load current of each of the plurality of power distribution systems A to C can be adjusted by opening and closing the sectional disconnectors 6.

[0023] Next, the opening and closing operation procedure of the sectional disconnector 6 and an example of the power distribution systems A to C after the opening and closing operation when performing a power outage operation on the power system 3 will be described with reference to Figs. 2 to 6. Specifically, taking the section 72 of the power system 3 in the state shown in Fig. 2 as the power outage target section X, an example of changing the power system 3 to each state in Figs. 3 to 6 in order to perform the power outage operation will be described.

[0024] Fig. 3 is a schematic diagram showing an example of each of the power distribution systems A to C when performing a power outage operation on the power system 3 at night during rainy weather. Fig. 4 is a schematic diagram showing an example of each of the power distribution systems A to C when performing a power outage operation on the power system 3 at night during rainy weather. Fig. 5 is a schematic diagram showing an example of each of the power distribution systems A to C when performing a power outage operation on the power system 3 during sunny weather in the daytime. Fig. 6 is a schematic diagram showing an example of each of the power distribution systems A to C when performing a power outage operation on the power system 3 during sunny weather in the daytime.

[0025] In the power system 3 in the states shown in FIGS. 2 to 6, the allowable load current of SVR83 is 350 A, and the allowable load current of the circuit breaker 82 is 200 A. That is, it is necessary to suppress the load current flowing from the distribution system B to the downstream side of SVR83 or from the distribution systems A and C to the downstream side of SVR83 to 350 A or less. Also, it is necessary to suppress the load current flowing from the distribution system C to the downstream side of the circuit breaker 82 to 200 A or less. Further, in the power system 3 in the states shown in FIGS. 3 to 6, the estimated value of the load current flowing through section 73 is 60 A, the estimated value of the load current flowing through section 74 is 60 A, the estimated value of the load current flowing through section 75 is 200 A, the estimated value of the load current flowing through section 76 is 25 A, and the estimated value of the load current flowing through section 77 is 5 A. The estimated power generation amount by the distributed power source 9 is 320 A during sunny days in the daytime and 0 A during rainy nights.

[0026] When the power outage target section X of the power system 3 in the state shown in FIG. 2 is powered off, not only section 72 but also the power supply to sections 73 and 74 downstream of section 72 as viewed from the first substation 5a stops. In order to avoid power outages in sections 73 and 74 that are not the power outage target section X, it is necessary to supply power to sections 73 and 74 from the distribution system B or the distribution system C.

[0027] First, in the example shown in FIG. 3, after closing the sectional disconnector 65 from the state shown in FIG. 2, the sectional disconnector 63 is opened. As a result, power is supplied from the distribution system B to sections 73 and 74. In the example of FIG. 3, since the power outage work is carried out during a rainy night, the distributed power source 9 connected to the distribution line 4 of section 73 does not generate power, and the estimated value of the load current on the downstream side of SVR83 in the distribution system B becomes 345 A. In this case, if the load in sections 73 to 77 increases by 5 A or more than assumed, it exceeds the allowable load current of SVR83, which is 350 A, and there is a risk of damage to SVR83.

[0028] On the one hand, in the example shown in FIG. 4, after closing the sectional disconnector 67 from the state shown in FIG. 2, the sectional disconnector 63 is opened. As a result, power is supplied from the power distribution system C to the sections 73 and 74. In the example shown in FIG. 4, since the power outage work is carried out at night during rainy weather, the distributed power source 9 connected to the distribution line 4 in the section 73 does not generate electricity, and the estimated value of the load current on the downstream side of the circuit breaker 82 in the power distribution system C is 125 A. On the other hand, since the allowable load current of the circuit breaker 82 is 200 A, even if the loads in the sections 73, 74, and 77 increase slightly more than expected, the allowable load current of the circuit breaker 82 will not be exceeded. That is, it can be said that the opening and closing operation procedure of the sectional disconnector 6 shown in the example of FIG. 4 is a more appropriate procedure compared to the opening and closing operation procedure shown in the example of FIG. 3.

[0029] In the example shown in FIG. 5, by the same opening and closing operation procedure as in the example shown in FIG. 4, without increasing or decreasing the section 7 supplied with power by the power distribution system B, the sections 73 and 74 are added as the sections supplied with power by the power distribution system C. However, different from the situation in FIG. 4, since the power outage work is carried out during sunny weather during the day, the distributed power source 9 connected to the section 73 generates electricity, and the estimated value of the load current on the downstream side of the circuit breaker 82 in the power distribution system C is -195 A. That is, in the power distribution system C, a reverse current of 195 A flows from the distributed power source 9 towards the circuit breaker 82. Since the allowable load current of the circuit breaker 82 is 200 A, there is a risk that the circuit breaker 82 will operate and the power supply will be cut off when the loads in the sections 73, 74, and 77 decrease more than expected or when the generated power of the distributed power source 9 increases more than expected.

[0030] On the one hand, in the example shown in FIG. 6, after the sectional disconnector 65 is switched on and the sectional disconnector 66 is opened from the state shown in FIG. 2, the sectional disconnector 67 is switched on and the sectional disconnector 63 is opened. As a result, the section 75 is disconnected from the section 7 supplied with power by the power distribution system B, and the sections 73, 74, and 75 are added as the sections 7 supplied with power by the power distribution system C. In the example of FIG. 6, since the power outage work is carried out on a sunny day during the day, the distributed power source 9 connected to the distribution line 4 in the section 73 generates power, and the estimated value of the load current of the power distribution system C is 5 A. Also, the estimated value of the load current of the power distribution system B is 25 A. On the other hand, the allowable load current of the circuit breaker 82 in the power distribution system C is 200 A, and the allowable load current of the SVR 83 in the power distribution system B is 350 A. Therefore, even if the load in the sections 73 to 77 increases slightly more than expected, it will not exceed the allowable load current. That is, it can be said that the opening / closing operation procedure of the sectional disconnector 6 shown in the example of FIG. 6 is a more appropriate procedure than the opening / closing operation procedure of the sectional disconnector 6 shown in the example of FIG. 5.

[0031] As shown in FIGS. 3 to 6, when the distributed power source 9 is connected to the distribution line 4, the load current flowing through each section 7 varies depending on the weather during the power outage work. Also, in the examples shown in FIGS. 3 to 6, the estimated values of the load currents during the day and at night in each section 7 are set to the same value, but generally, the load current varies greatly depending on the time zone. For this reason, it is important to select an appropriate pattern of opening / closing operation procedures according to the load current that changes according to various factors such as the time zone and the weather.

[0032] Regarding this point, the opening / closing operation procedure determination system 100 of the present embodiment is devised for the purpose of enabling an operator to more efficiently determine the opening / closing operation procedure of the sectional disconnector 6 that more accurately reflects the load current in each section 7 during the power outage work.

[0033] Next, the configuration of the opening / closing operation procedure determination system 100 will be described. As shown in FIG. 1, the opening / closing operation procedure determination system 100 includes a machine learning device 1 and an opening / closing operation procedure determination device 2.

[0034] Here, the machine learning device 1 and the opening / closing operation procedure determination device 2 are paired one-to-one and are communicably connected. Although not shown in FIG. 1, the machine learning device 1 and the opening / closing operation procedure determination device 2 may be connected to each other via a network. The network is, for example, a LAN (Local Area Network), the Internet, a public telephone network, or a combination thereof. The specific communication method in the network, whether it is a wired connection or a wireless connection, etc. is not particularly limited. Alternatively, the machine learning device 1 and the opening / closing operation procedure determination device 2 may be directly connected via a connector instead of communicating using a network.

[0035] The machine learning device 1 will be described with reference to FIG. 7. FIG. 7 is a diagram showing the functional blocks of the machine learning device 1.

[0036] The machine learning device 1 constructs a learning model used in the opening / closing operation procedure determination device 2 by supervised learning. The learning model constructed by the machine learning device 1 is used to determine the opening / closing operation procedure of the sectionalizer 6 during a power outage operation in the power system 3. As shown in FIG. 3, the machine learning device 1 includes an input data acquisition unit 11, a label acquisition unit 12, a storage unit 13, and a learning model construction unit 14.

[0037] The input data acquisition unit 11 acquires various types of information related to the power outage operation of the power system 3 as input data. The input data acquired by the input data acquisition unit 11 includes a data group including power system information, allowable load current, power outage target section X, time zone of the power outage operation, weather during the power outage operation, load current of each section 7 during the time zone of the power outage operation, and power generation amount of the distributed power source 9 arranged in each section 7 during the time zone of the power outage operation.

[0038] The power system information is information regarding the power system 3 where the power outage work is to be performed. Examples of the power system information include the location information of the substation 5, the location information of the distribution line 4, and the location information of the equipment arranged between a plurality of substations 5 and partitioning the distribution line 4. Examples of the location information of the equipment partitioning the distribution line 4 include the location information of a plurality of sectional switches 6, the location information of circuit breakers 81, 82, and SVR 83, etc. The input data acquisition unit 11 may acquire the power system information from a management system of the power system 3 such as a distribution automation system, for example.

[0039] The allowable load current is the allowable value of the load current set to protect equipment such as the distribution line 4 within the power system 3. Examples of the allowable load current include the allowable load current of each of distribution systems A to C, the allowable load current of the circuit breaker 81, the allowable load current of the circuit breaker 82, the allowable load current of the SVR 83, etc. The input data acquisition unit 11 may acquire information regarding the allowable load current from a management system of the power system 3 such as a distribution automation system, for example.

[0040] The power outage target section X is the location information of the section 7 that is the target of the power outage work. For example, in the examples shown in FIGS. 3 to 6, the section 72 becomes the power outage target section X. The input data acquisition unit 11 may acquire, as the power outage target section X, for example, information input by an operator via an input device provided in the machine learning device 1.

[0041] The time zone of the power outage work may be classified into morning, noon, night, etc., or may be classified at predetermined time intervals. The input data acquisition unit 11 may acquire, as the time zone of the power outage work, information input by an operator via an input device provided in the machine learning device 1.

[0042] The weather during the power outage work is information regarding the weather of the power system 3 during the power outage work. Examples of the weather during the power outage work include information such as sunny, rainy, cloudy, etc. The input data acquisition unit 11 may be, for example, information input by an operator via an input device provided in the machine learning device 1, or may acquire the announcement of a meteorological observation station as the weather information.

[0043] As the load current in each section 7 during the power outage operation time period, for example, it may be the load current in each section 7 during the power outage operation, or it may be the load current in each section 7 in the same time period in the past rather than during the power outage operation. Also, the load current in each section 7 may be an actual measurement value, an estimated value, or an average value of a plurality of actual measurement values. The input data acquisition unit 11 may acquire information regarding the load current, for example, from a database provided in a management system of the power grid 3 such as a distribution automation system, or may acquire it from smart meters of each load within the power grid 3.

[0044] The power generation amount by the distributed power source 9 in each section 7 during the time period of the power outage time may be the power generation amount by the distributed power source 9 in each section 7 during the power outage operation, or it may be the power generation amount by the distributed power source 9 in each section 7 in the same time period in the past rather than during the power outage operation and under the same weather conditions. Also, the power generation amount by the distributed power source 9 in each section 7 may be an actual measurement value or an estimated value. The input data acquisition unit 11 may acquire information regarding the power generation amount of the distributed power source 9 in each section 7, for example, from a database provided in a management system of the power grid 3 such as a distribution automation system or from a distributed power source ledger.

[0045] The label acquisition unit 12 acquires, as a label, the opening / closing operation procedure of the sectional disconnector 6 for adjusting the load current of each of the distribution systems A to C during the power outage operation. The opening / closing operation procedure of the sectional disconnector 6 is information regarding the procedure of the opening / closing operation of the sectional disconnector 6, that is, which sectional disconnector 6 among a plurality of sectional disconnectors 6 is switched on and off at which timing. The opening / closing operation procedure of the sectional disconnector 6 is preferably, for example, an operation procedure in which the ratio of the load current to the allowable load current set within the power grid 3 becomes smaller. For example, in the power grid 3, the opening / closing operation procedure of the sectional disconnector 6 is preferably a procedure for adjusting the load current of the distribution systems A to C so that the ratio of the load current to the allowable load current of each of the distribution systems A to C and the circuit breakers 81, 82, and SVR 83 becomes smaller. The label acquisition unit 12 may acquire, as a label, the opening / closing operation procedure of the sectional disconnector 6 during past power outage operations from the management system of the power grid 3, for example.

[0046] The storage unit 13 stores, as a set of data groups, the power system information acquired by the input data acquisition unit 11, the allowable load current, the power outage target section X, the time zone of the power outage work, the weather during the power outage work, the load current of each section 7, and the power generation amount by the distributed power source 9 arranged in each section 7, in association with each other. Then, the storage unit 13 stores, as a set of teacher data, the input data which is a set of data groups and the label which is the opening / closing operation procedure of the sectional disconnector 6 for the input data, in association with each other. Further, the storage unit 13 also stores the learning model constructed by the learning model construction unit 14.

[0047] The learning model construction unit 14 constructs a learning model for determining the opening / closing operation procedure of the sectional disconnector 6 for adjusting the load current of each of the distribution systems A to C during the power outage work, by performing supervised learning using the set of the input data and the label stored in the storage unit 13 as teacher data. Then, the learning model construction unit 14 transmits the constructed learning model to the opening / closing operation procedure determination device 2.

[0048] The learning model construction unit 14 can be realized, for example, by using a Support Vector Machine (hereinafter also referred to as SVM).

[0049] Here, the loads connected to the distribution line 4 of the power system 3 have different power usage situations depending on their types. For example, when the load is a factory, the power usage amount is large compared to other loads, and there is a tendency for the power usage amount to peak during the daytime on weekdays. In the case of a store or an office, the power usage amount peaks during the daytime, and in the case of a residence, the power usage amount peaks in the morning and evening. That is, the load current flowing through each section 7 varies according to the time zone, and the magnitude of the load current and the variation pattern of the daily load current also differ according to the type of load existing in each section 7. Also, the human behavior pattern changes depending on the weather, and the power usage situations of stores, residences, etc. change.

[0050] In this embodiment, since information associating the time period of the power outage work, the weather, and the load current in each section 7 is learned as teacher data, a learning model that more accurately reflects the actual results of the load current in each section 7 can be constructed.

[0051] Also, since the power generation amount by the distributed power source 9 changes depending on the solar radiation amount, the weather has a great influence on the load current in each section 7. In this embodiment, since the power generation amount by the distributed power source 9 in each section 7 is also learned as teacher data together with the weather, a learning model with higher accuracy can be constructed.

[0052] Also, in this embodiment, not only the allowable load currents of the distribution systems A to C but also the allowable load currents of SVR83 etc. provided in the middle of the distribution path are used as teacher data. From this, a learning model for more reliably determining the switching operation procedure in which the load current of each of the distribution systems A to C does not exceed each allowable load current set in the power system 3 during the power outage work can be constructed.

[0053] The switching operation procedure determination device 2 will be described with reference to FIG. 8. FIG. 8 is a functional block diagram of the switching operation procedure determination device 2. The switching operation procedure determination device 2 includes a control unit 20, a storage unit 24, a communication unit 25, and a display unit 26.

[0054] The control unit 20 is a part that controls the entire switching operation procedure determination device 2, and realizes various functions in this embodiment by appropriately reading and executing various programs from a storage area such as a ROM, a RAM, a flash memory, or a hard disk drive (HDD). The control unit 20 may be a CPU. The control unit 20 includes a determination data acquisition unit 21, a switching operation procedure determination unit 22, and an output unit 23.

[0055] The determination data acquisition unit 21 acquires determination data for the opening / closing operation procedure determination unit 22 to determine the opening / closing operation procedure of the sectionalizer 6 during a power outage operation. The determination data acquisition unit 21 includes a power grid information acquisition unit 211, an allowable load current information acquisition unit 212, a power outage target section information acquisition unit 213, a time zone information acquisition unit 214, a weather information acquisition unit 215, a load current information acquisition unit 216, and a distributed power source information acquisition unit 217.

[0056] The power grid information acquisition unit 211 acquires, as determination data, power grid information that is information regarding the power grid 3 where the power outage operation is performed. The power grid information acquired by the power grid information acquisition unit 211 may be information included in the teacher data used during machine learning, or may be the latest information regarding the current power grid 3 acquired separately from the information included in the teacher data.

[0057] The allowable load current information acquisition unit 212 acquires, as determination data, the allowable load current within the power grid 3, such as the allowable load current of each of the distribution grids A to C, the allowable load current of the circuit breakers 81 and 82, and the allowable load current of the SVR 83. The allowable load current acquired by the allowable load current information acquisition unit 212 may be information included in the teacher data used during machine learning, or may be the latest allowable load current at the current time acquired separately from the information included in the teacher data.

[0058] The power outage target section information acquisition unit 213 acquires, as determination data, the power outage target section X that is the target of a new power outage operation.

[0059] The time zone information acquisition unit 214 acquires, as determination data, the time zone of a new power outage operation.

[0060] The weather information acquisition unit 215 acquires, as determination data, the weather during a new power outage operation. The weather information acquisition unit 215 may be information obtained from the announced values of a meteorological observation station, for example.

[0061] The load current information acquisition unit 216 acquires, as determination data, the load current of each section 7 during the time period of the new power outage work. The load current of each section 7 may be information included in the teacher data used during machine learning, or may be information acquired separately from the information included in the teacher data.

[0062] The distributed power source information acquisition unit 217 acquires, as determination data, the power generation amount of the distributed power source 9 in each section 7 during the time period of the new power outage work. The power generation amount by the distributed power source 9 in each section 7 may be information included in the teacher data used during machine learning, or may be information acquired separately from the information included in the teacher data.

[0063] The opening / closing operation procedure determination unit 22 determines the opening / closing operation procedure of the sectional disconnector 6 during the power outage work of the power grid 3 based on the determination data acquired by the determination data acquisition unit 21 and the learning model.

[0064] The output unit 23 outputs the opening / closing operation procedure of the sectional disconnector 6 determined by the opening / closing operation procedure determination unit 22 to the display unit 26.

[0065] The storage unit 24 stores the learning model acquired from the machine learning device 1. Further, the storage unit 24 sets the power grid information acquired by the determination data acquisition unit 21, the allowable load current in the power grid 3, the power outage target section X, the time period of the power outage work, the weather during the power outage work, the load current of each section 7 during the same time period as the power outage work, and the power generation amount by the distributed power source 9 in each section 7 during the same time period as the power outage work as a set of input data, and may store the opening / closing operation procedure of the sectional disconnector 6 determined for the input data as a label.

[0066] The communication unit 25 transmits and receives data to and from the machine learning device 1. For example, the opening / closing operation procedure determination system 100 can transmit the learning model constructed by the learning model construction unit 14 from the machine learning device 1 to the opening / closing operation procedure determination device 2 via the communication unit 25. Conversely, the opening / closing operation procedure determination system 100 can transmit, from the opening / closing operation procedure determination device 2 to the machine learning device 1 via the communication unit 25, new power system information, allowable load current, power outage target section X, time zone of the power outage work, weather during the power outage work, load current of each section 7, power generation amount of the distributed power source 9 in each section 7, the determined opening / closing operation procedure, etc., acquired by the determination data acquisition unit 21.

[0067] The display unit 26 is a monitor that displays the determination result of the opening / closing operation procedure of the sectional disconnector 6 output from the output unit 23 of the control unit 20.

[0068] Next, the operation during machine learning in the opening / closing operation procedure determination system 100 according to the present embodiment will be described. FIG. 9 is a flowchart showing the operation of the machine learning device 1 during this machine learning.

[0069] In step S11, the input data acquisition unit 11 of the machine learning device 1 acquires, as input data, power system information, allowable load current, power outage target section X, time zone of the power outage work, weather during the power outage work, load current of each section 7 in the time zone of the power outage work, and power generation amount by the distributed power source 9 in each section 7 in the time zone of the power outage work.

[0070] In step S12, the label acquisition unit 12 of the machine learning device 1 acquires the opening / closing operation procedure of the sectional disconnector 6 during the power outage work as a label.

[0071] In step S13, the learning model construction unit 14 of the machine learning device 1 receives the set of the input data and the label as teacher data.

[0072] In step S14, the learning model construction unit 14 of the machine learning device 1 executes machine learning using this teacher data.

[0073] In step S15, the learning model construction unit 14 determines whether to end the machine learning or repeat the machine learning. If the learning model construction unit 14 determines to repeat the machine learning (No in step S15), the process returns to step S11. Then, the machine learning device 1 repeats the same operation. On the other hand, if the learning model construction unit 14 determines to end the machine learning (Yes in step S15), the process proceeds to step S16. Note that the conditions for ending the machine learning can be arbitrarily determined. For example, the machine learning may be ended when the machine learning is repeated a predetermined number of times.

[0074] In step S16, the machine learning device 1 transmits the learning model constructed by the machine learning up to that point to the opening / closing operation procedure determination device 2 via a network or the like.

[0075] Also, the storage unit 13 of the machine learning device 1 stores this learning model. Thereby, when the opening / closing operation procedure determination device 2 requests the learning model, the learning model can be transmitted to the opening / closing operation procedure determination device 2. Also, when new teacher data is acquired, further machine learning can be performed on the learning model.

[0076] Next, an example of the process of determining the opening / closing operation procedure by the opening / closing operation procedure determination device 2 will be described with reference to FIG. 10. FIG. 10 is a flowchart showing the flow of the process of determining the opening / closing operation procedure of the sectionalizer 6 during a power outage operation by the opening / closing operation procedure determination device 2.

[0077] In step S21, the determination data acquisition unit 21 acquires determination data including power system information, the allowable load current set in the power system 3, the power outage target section X, the time zone of the power outage operation, the weather during the power outage operation, the load current of each section 7 in the same time zone as during the power outage operation, and the power generation amount by the distributed power source 9 arranged in each section 7 in the same time zone as during the power outage operation.

[0078] In step S22, the opening / closing operation procedure determination unit 22 determines the opening / closing operation procedure of the sectional disconnector 6 based on the power system information acquired in step S21, the allowable load current set in the power system 3, the power outage target section X, the time zone of the power outage work, the weather during the power outage work, the load current of each section 7 in the same time zone as during the power outage work, a set of data including the power generation amount by the distributed power source 9 arranged in each section 7 in the same time zone as during the power outage work, and the learning model transmitted from the machine learning device 1.

[0079] In step S23, the output unit 23 outputs the opening / closing operation procedure of the sectional disconnector 6 determined in step S22 to the display unit 26. As a result, the opening / closing operation procedure of the sectional disconnector 6 determined by the opening / closing operation procedure determination unit 22 is displayed on the display unit 26.

[0080] According to the machine learning device 1 or the opening / closing operation procedure determination device 2 of the opening / closing operation procedure determination system 100 according to the present embodiment described above, the following effects can be obtained.

[0081] The machine learning device 1 according to the present embodiment includes an input data acquisition unit 11 that acquires, as input data, a data group including the power outage target section X in the power system 3 where the load current of each of the plurality of power distribution systems A to C can be adjusted by opening and closing the sectional disconnector 6 that divides the distribution line 4 into a plurality of sections 7, the time zone of the power outage work for powering off the power outage target section X, the weather during the power outage work, and the load current of each of the plurality of sections 7 in the time zone; a label acquisition unit 12 that acquires, as a label, the opening / closing operation procedure of the sectional disconnector 6 for adjusting the load current of each of the plurality of power distribution systems A to C during the power outage work; and a learning model construction unit 14 that constructs a learning model for determining the opening / closing operation procedure of the sectional disconnector 6 during the power outage work by performing supervised learning using the set of the input data and the label as the training data.

[0082] Thereby, a learning model for more efficiently determining the opening / closing operation procedure of the sectional disconnector 6 during the power outage work in consideration of the time zone, weather, etc. of the power outage work can be constructed.

[0083] In the machine learning device 1 according to the present embodiment, the input data acquisition unit 11 further acquires, as input data, the power generation amount by the distributed power source 9 arranged in each of the plurality of sections 7 in the time zone.

[0084] Thereby, it is possible to determine the opening / closing operation procedure of the sectional disconnector 6 that more accurately reflects the power generation amount by the distributed power source 9.

[0085] Further, the opening / closing operation procedure determination device 2 according to the present embodiment is an opening / closing operation procedure determination device 2 that uses the learning model constructed by the machine learning device 1, and includes a new power outage target section X in the power grid 3, a time zone of a new power outage operation for powering off the new power outage target section X, the weather during the new power outage operation, and load currents of each of the plurality of sections 7 in the time zone of the new power outage operation. A determination data acquisition unit 21 that acquires determination data including the above, and an opening / closing operation procedure determination unit 22 that determines the opening / closing operation procedure of the sectional disconnector 6 during the new power outage operation based on the determination data and the learning model.

[0086] Thereby, by inputting the time zone of the new power outage operation, the weather during the power outage operation, the power outage target section X, etc. into the learning model of the present embodiment, the opening / closing operation procedure of the sectional disconnector 6 is determined. Therefore, regardless of the number of years of experience and the amount of knowledge of the operator, an opening / closing operation procedure considering past performance can be efficiently obtained.

[0087] As described above, the embodiments of the present invention have been described, but the present invention is not limited to the above embodiments and can be appropriately changed.

[0088] In the above embodiment, the machine learning device 1 has acquired a data group including power grid information and allowable load current as input data and performed supervised learning. However, a configuration may be adopted in which supervised learning is performed without acquiring power grid information and allowable load current as input data. In addition, when performing supervised learning without acquiring power grid information and allowable load current, a highly accurate learning model can be constructed by repeating machine learning for the same power grid 3.

[0089] Further, the opening / closing operation procedure determination device 2 may be configured not to include the power system information acquisition unit 211 and the allowable load current information acquisition unit 212. In this case, the opening / closing operation procedure determination device 2 may be configured to determine the opening / closing operation procedure of the sectionalizer 6 based on the determination data that does not include the power system information and the allowable load current, and the learning model constructed without inputting the power system information and the allowable load current.

Explanation of Signs

[0090] 1 Machine learning device 2 Opening / closing operation procedure determination device 3 Power system 4 Distribution line 6, 61, 62, 63, 64, 65, 66, 67 Sectionalizer 7, 71, 72, 73, 74, 75, 76, 77 Section 11 Input data acquisition unit 12 Label acquisition unit 14 Learning model construction unit A, B, C Distribution system

Claims

1. An input data acquisition unit that acquires, as input data, a data group including a power outage target section in a power system in which the load current of each of a plurality of power distribution systems can be adjusted by opening and closing sectionalizers that divide a distribution line into a plurality of sections, a time zone of a power outage operation for powering off the power outage target section, the weather during the power outage operation, and the load current of each of the plurality of sections in the time zone; A label acquisition unit that acquires, as a label, an opening / closing operation procedure of the sectionalizer for adjusting the load current of each of the plurality of power distribution systems during the power outage operation; A machine learning device comprising a learning model construction unit that constructs a learning model for determining the opening / closing operation procedure of the sectionalizer during the power outage operation by performing supervised learning using a set of the input data and the label as supervised data.

2. The machine learning device according to claim 1, wherein the input data acquisition unit further acquires, as input data, the power generation amount by distributed power sources arranged in each of the plurality of sections in the time zone.

3. An opening / closing operation procedure determination device using the learning model constructed by the machine learning device according to claim 1 or 2, A determination data acquisition unit that acquires determination data including a new power outage target section in the power system, a time zone of a new power outage operation for powering off the new power outage target section, the weather during the new power outage operation, and the load current of each of the plurality of sections in the time zone of the new power outage operation; An opening / closing operation procedure determination device comprising an opening / closing operation procedure determination unit that determines the opening / closing operation procedure of the sectionalizer during the new power outage operation based on the determination data and the learning model.

Images