JP2022085363A - Predictive model generation device and method - Google Patents

Abstract

To provide a predictive model generation device and method that can improve the quality of a distribution service and eliminate the busy feeling of responding to power outages while reducing the cost required for maintenance of a distribution system.SOLUTION: In a predictive model generation method executed by a predictive model generation device that generates a predictive model for predicting the occurrence of equipment defects below a the low-voltage drop line, on the basis of the frequency of occurrence of each specific event recognized on the basis of event information of each specific event detected by a smart meter installed in each customer, specific events detected by the smart meter within a predetermined first period are respectively leveled to a plurality of levels, the number of times a specific event of each level has occurred is counted for each specific event, and a predictive model is generated on the basis of the number of occurrences of each specific event counted for each level.SELECTED DRAWING: Figure 4

Description

The present invention relates to a predictive model generator and a method, and is suitable for application to, for example, a predictive model generator for generating a predictive model for predicting the occurrence of equipment defects below a low voltage drop line in a distribution system.

In recent years, smart meters, which are electric energy meters equipped with two-way communication functions, have been installed in consumers. By using a smart meter, detailed data such as the amount of electricity used by consumers every 30 minutes (hereinafter referred to as the 30-minute value) and reverse power flow from private power generation, which could not be obtained by the conventional weighing system, can be obtained. , Information on various events such as voltage drop, power failure and power recovery detected by the smart meter (hereinafter referred to as event information) can be acquired.

Patent Document 1 discloses a technique for collecting periodic meter reading values, which is a basic function of such a smart meter. In recent years, various technologies related to smart meters such as technologies related to other communication methods have been proposed.

Furthermore, in recent years, with the spread of smart meters, there is a movement to consider the utilization of meter reading value information such as 30-minute values and monthly power consumption transmitted from smart meters to the center equipment side, and event information of various events. There is also. For example, in Patent Document 2, the cause of an accidental power outage or a failure of a power distribution facility is specified in detail by utilizing a 30-minute value or event information from a smart meter installed in a power pipe, and then the specified cause is specified. A data management system has been proposed that can instruct how to deal with these accidental power outages and power distribution equipment failures.

Further, Patent Document 3 discloses a technique for estimating the position where a failure such as disconnection occurs based on the event notification of the smart meter in relation to the phase group estimation processing program for estimating the failure position of the power distribution equipment. ..

Japanese Unexamined Patent Publication No. 2005-352532 Japanese Unexamined Patent Publication No. 2018-207690 International Publication No. 2018/179283

By the way, in general, it is difficult to predict equipment defects below the low-voltage drop line, such as a blown fuse of a fuse provided in a low-voltage drop line connecting a transformer installed on a utility pole and a smart meter, or a broken wire. For this reason, in the maintenance management work of the distribution system, there are problems such as an increase in cost due to emergency commissioned work, deterioration of service quality due to the occurrence of a defect in the low voltage drop line, and a feeling of busyness in dealing with power outages day and night.

Further, in the techniques disclosed in Patent Document 2 and Patent Document 3, since a program dedicated to the relevant equipment is required to identify and estimate the failure of the electric power equipment, it is not easy to expand to other equipment.

The present invention has been made in consideration of the above points, and solves such problems at once, reduces the cost required for maintenance and management of the distribution system, improves the quality of the distribution service, and is busy responding to power outages. We are trying to propose a predictive model generation device and a method that can eliminate the feeling.

In order to solve such a problem, in the present invention, in the prediction model generator that generates a prediction model for predicting the occurrence of equipment defects below the low-voltage drop line, each identification detected by the smart meter installed in each customer. Based on the frequency of occurrence of each specific event recognized based on the event information of the event, each specific event detected by the smart meter within a predetermined first period is leveled to a plurality of levels, respectively. The prediction model is based on an event leveling unit that counts the number of times the specific event of the specific event has occurred for each specific event, and the number of occurrences of each specific event for each level counted by the event leveling unit. A predictive model generation unit to be generated is provided.

Further, the present invention is a predictive model generation method executed by a predictive model generator that generates a predictive model for predicting the occurrence of equipment defects below the low-pressure drop line, and is a smart meter installed in each customer. Based on the frequency of occurrence of each specific event recognized based on the event information of each specific event detected by the smart meter, each specific event detected by the smart meter is divided into multiple levels within a predetermined first period. The prediction model is based on the first step of leveling each and counting the number of occurrences of the specific event of each said level for each specific event, and the number of occurrences of each of the counted specific events for each level. A second step to generate is provided.

According to the predictive model generation device and method of the present invention, the occurrence of equipment defects below the low-pressure drop line can be predicted with high accuracy based on the generated prediction model. By taking advance measures, it is possible to increase the cost of maintenance management work of the distribution system due to emergency commissioned work and reduce the frequency of power outage response.

According to the present invention, it is possible to realize a predictive model generation device and a method capable of improving the quality of a distribution service and eliminating the busy feeling of responding to a power outage while reducing the cost required for maintenance and management of the distribution system.

It is a block diagram which shows the physical composition of the prediction model generation apparatus by this embodiment. It is a figure which shows the structure of the event information management table. It is a chart which shows the structure of a 30-minute value management table. It is a block diagram which shows the logical structure of the prediction model generation apparatus by this embodiment. (A) and (B) are diagrams and charts used for explaining the leveling of a specific event by the event leveling unit. It is a figure which provides the explanation of the leveling of a specific event by the event leveling part. It is a chart provided for explanation of event leveling information. It is a chart used for explanation of learning data. It is a figure which shows the structural example of the prediction model. It is a figure which provides the explanation of the learning data acquisition period, the verification data acquisition period, and the actual data acquisition period. It is a graph which provides the explanation of the addition period of the verification data acquisition period. (A) to (C) are Venn diagrams used for explaining the hit rate and the coverage rate. It is a figure which shows the configuration example of the 1st event number management table. It is a figure which shows the configuration example of the event leveling table. It is a figure which shows the configuration example of the power consumption management table. It is a figure which shows the configuration example of the equipment information management table. It is a figure which shows the configuration example of the elapsed time management table.

Hereinafter, one embodiment of the present invention will be described in detail with reference to the drawings.

(1) Configuration of Predictive Model Generation Device According to the Present Embodiment (1-1) Physical Configuration of Predictive Model Generation Device In FIG. 1, 1 shows a prediction model generation device according to the present embodiment. The predictive model generation device 1 includes a restriction cancellation discount system 2, a meter data management system (MDMS: Meter Data Management System) 3, and various other systems 4.

The prediction model generation device 1 has a prediction model generation function for generating a prediction model for predicting the occurrence of equipment defects (hereinafter referred to as power outages) below the low-voltage service line in the distribution system, and the CPU (Central Processing Unit) 10 , A general-purpose computer device including a memory 11A, a storage device 11B, and a display device 12.

The CPU 10 is a processor that controls the operation of the entire predictive model generation device 1. Further, the memory 11A is composed of, for example, a volatile semiconductor memory and is used as a work memory of the CPU 10. The memory 11A stores an analysis tool 13 described later, which is application software for realizing such a predictive model generation function.

Further, the storage device 11B is used for holding a program and necessary information for a long period of time, and is composed of a hard disk device and a large-capacity non-volatile storage device such as an SSD (Solid State Drive) or a flash memory. In addition to various programs, the storage device 11B has a first event number management table 14, an event leveling table 15, a power consumption management table 16, a facility information management table 17, an elapsed time management table 18, and a second event, which will be described later. The number of times management table 19 and the prediction model 20 generated by the prediction model generation function are stored.

The display device 12 is composed of, for example, a liquid crystal display or an organic EL (Electro Luminescence) display, and is used for displaying necessary information.

On the other hand, the meter data management system 3 is a computer device having a function of managing event information and 30-minute values transmitted from each smart meter (not shown) in the jurisdiction, and configuration information of each smart meter. ..

In practice, the meter data management system 3 registers and manages event information appropriately transmitted from smart meters installed in each consumer in an event information management table 21 as shown in FIG. In FIG. 2, the information in each row of the event information management table 21 represents the event information 40, respectively.

Further, the meter data management system 3 registers and manages the 30-minute values of the corresponding consumers transmitted from each smart meter, for example, in a 30-minute cycle, in the 30-minute value management table 22 as shown in FIG. .. In FIG. 3, the value in each column of each row of the 30-minute value management table 22 represents the 30-minute value 42 in the corresponding time zone of one consumer. Further, the meter data management system 3 registers and manages the configuration information of each smart meter such as a single-phase three-wire type or a three-phase three-wire type in a configuration information table (not shown).

These information (event information 40, consumer 30-minute value 42, and smart meter configuration information) managed by the meter data management system 3 is, for example, a semiconductor memory such as a USB (Universal Serial Bus) memory. It is manually provided to the prediction model generation device 1 via an optical disk such as a CD (Compact Disc) or DVD (Digital Versatile Disc) or a portable storage device such as a portable hard disk device.

The restriction cancellation discount system 2 is composed of one or a plurality of computer devices that manage the results of various power outages that have occurred within the jurisdiction. Further, the various systems 4 are composed of one or a plurality of computer devices that manage the installation date of each smart meter to the consumer, inspection results, equipment design information, completion information, and equipment information such as repair results.

Information on the actual results of various power outages managed by the limited cancellation discount system 2 and the above-mentioned equipment information managed by the various systems 4 are also manually predicted model generators via semiconductor memory, optical disk or portable storage device. Provided in 1.

(1-2) Logical Configuration of Predictive Model Generator FIG. 4 shows a logical configuration of the predictive model generator 1. As is clear from FIG. 4, the prediction model generation device 1 includes the first event frequency management unit 30, the event leveling unit 31, the power consumption management unit 32, the equipment information management unit 33, and the elapsed time management unit 34. Second event number management unit 35, prediction model generation unit 36, model evaluation unit 37, defect occurrence prediction unit 38, first event number management table 14, event leveling table 15, power consumption management table 16, equipment information It is configured to include a management table 17, an elapsed time management table 18, and a second event count management table 19.

First event number management unit 30, event leveling unit 31, power consumption management unit 32, equipment information management unit 33, elapsed time management unit 34, second event number management unit 35, prediction model generation unit 36, model evaluation The unit 37 and the defect occurrence prediction unit 38 are all functional units embodied by the CPU 10 (FIG. 1) executing the analysis tool 13 (FIG. 1) stored in the memory 11A (FIG. 11).

The first event frequency management unit 30 is the predetermined period regarding the date and time when the equipment failure (power outage) occurred in any of the low-voltage service lines and the generated low-voltage service line from the information on the results of various power outages provided as described above. The power outage record information 45 for (at least the period from the present time in FIG. 10 to "a / aa") and the event information 40 for the same period are extracted from the provided event information 40, and these information are extracted. Based on the above, it has a function of counting the number of occurrences of each specific event (hereinafter referred to as a specific event) detected by each smart meter within that period for each smart meter and each event type. The first event count management unit 30 stores and manages the count result in the first event count management table 14, which will be described later with reference to FIG. 13.

Since the event information 40 acquired by the first event number management unit 30 generates the prediction model 20 based on the event information 40 as described later, an event corresponding to the event information 40 has occurred thereafter. It shall be a sign of equipment failure below the low pressure drop line. Therefore, it is essential that the equipment failure (power outage) has already occurred at this point. Therefore, when acquiring the event information 40, the first event number management unit 30 acquires the event information of the event that occurred in a predetermined period (hereinafter, referred to as a learning data acquisition period) to some extent before the present time. (See FIG. 10). The length of this learning data acquisition period is a period that includes many of the specific events (more than a predetermined ratio) that are signs of equipment failure below the low-voltage drop line that occurred at the end of the learning data acquisition period, and is empirically determined. The length is determined.

Further, in the following, it is assumed that the low voltage drop wire is a single-phase three-wire system. Therefore, in the event of each smart meter for which the event information 40 should be acquired by the first event count management unit 30, the voltage of the one-side electric wire of the low-voltage service line connected to the one-side terminal of the smart meter is lowered. "Drop", "1 side power failure" where the power supply from the 1 side electric wire is stopped, and the voltage of the 3 side electric wire of the low voltage drop line connected to the 3 side terminal of the smart meter is decreased or from the 3 side electric wire. There are three types: "three-side voltage drop" when the power supply is stopped.

However, in the present embodiment, "cover opening / closing" in which the cover of the smart meter is opened / closed by a worker or the like for inspection work is also an event for which event information 40 should be acquired. Even in this case, the first event frequency management unit 30 describes only the three types of events of "1 side voltage drop", "1 side power failure", and "3 side voltage drop" excluding "cover opening / closing". The number of occurrences is counted as a "specific event".

Based on the event information 40 acquired by the first event number management unit 30, the event leveling unit 31 sets each specific event within the learning data acquisition period detected by the smart meter from "1" to "1" for each smart meter. It has a function of leveling to 5 levels up to "5".

Specifically, as shown in FIGS. 5 (A) and 5 (B) and FIG. 6, the event leveling unit 31 is a specific event for which the smart meter is a leveling target (hereinafter, appropriately referred to as a leveling target specific event). When the smart meter detects a specific event of the same type 0 to 1 times within the immediately preceding predetermined time (“N time” in FIG. 6, for example, 3 hours). Levels the leveling target specific event to "level 1 (Lv.1)".

Further, the event leveling unit 31 sets the "level" when the number of times the smart meter detects a specific event of the same type within a predetermined time immediately before the smart meter detects the leveling target specific event is 2 to 4 times. 2 (Lv.2) ”,“ Level 3 (Lv.3) ”for 5 to 8 times,“ Level 4 (Lv.4) ”for 9 to 13 times, 14 times or more Levels the leveling target specific event to "Level 5 (Lv.5)" respectively.

Further, the event leveling unit 31 similarly levels the “all specific events” and the ““ 1-side power failure ”or“ 1-side voltage drop ””, respectively.

Specifically, the event leveling unit 31 identifies one of the "all specific events" within the predetermined time immediately before the smart meter detects any specific event (leveling target specific event). "Level 1 (Lv.1)" when the number of times the event is detected is 0 to 1, "Level 2 (Lv.2)" when the number of times is 2 to 4, and "Level 2 (Lv.2)" when the number of times the event is detected is 5 to 8 times. "Level 3 (Lv.3)", "Level 4 (Lv.4)" for 9 to 13 times, and "Level 5 (Lv.5)" for 14 times or more. Level each.

Further, in the event leveling unit 31, the smart meter detects a specific event (leveling target specific event) of "1 side power failure" or "1 side voltage drop" for "" 1 side power failure "or" 1 side voltage drop "". If the number of times the smart meter detects a specific event of "1 side power failure" or "1 side voltage drop" within the specified time immediately before the time is 0 to 1, "Level 1 (Lv.1)" "Level 2 (Lv.2)" for 2 to 4 times, "Level 3 (Lv.3)" for 5 to 8 times, and "Level 4 (Lv.3)" for 9 to 13 times. Lv.4) ”, and in the case of 14 times or more, level the leveling target specific event to“ level 5 (Lv.5) ”.

It should be noted that the reason why "" 1-side power failure "or" 1-side voltage drop "" is targeted for leveling is that the smart meter treats the power failure and voltage drop that occurred at the 1-side terminal as "1 side power failure" and "1 side voltage drop", respectively. While it can be detected and output as a different event called "3 side voltage drop", considering that the power failure and voltage drop that occurred in the 3 side terminal can be detected only as one type of event called "3 side voltage drop", the 1 side terminal and 3 This is to align the output units of the side terminals.

Then, the event leveling unit 31 uses the event information 40 as shown in FIG. 2 acquired by such leveling to be "one-side power failure", "one-side voltage drop", "three-side voltage drop", and "all specific events". And for each of "1 side power failure" or "1 side voltage drop", the corresponding specific event ("1 side power failure", "1 side voltage drop", "3 side voltage drop", "all specific events" or The event leveling information 41 as shown in FIG. 7 summarizing only the leveling results of "" 1-side power failure "or" 1-side voltage drop "") is processed.

As is clear from FIG. 7, the event leveling information 41 includes "one-side power failure", "one-side voltage drop", "three-side voltage drop", "all specific events", and "one-side power failure" or "one-side power failure". For each of "1 side voltage drop", it is information in which the number of times the corresponding specific event has occurred within a predetermined time immediately before the occurrence and the level given to the specific event by the above-mentioned leveling are associated with each other.

Further, the event leveling unit 31 has "1 side power failure", "1 side voltage drop", and "3 side voltage drop" detected by the smart meter for each smart meter based on the event leveling information 41 obtained by such processing. , "All specific events" and "1 side power failure" or "1 side voltage drop" are counted for each level, and the count results are counted as "1 side power failure" and "1", which will be described later with reference to FIG. 14, respectively. It is stored and managed in the event leveling table 15 generated for each of "side voltage drop", "three-side voltage drop", "all specific events" and "" one-side power failure "or" one-side voltage drop "".

The power consumption management unit 32 extracts the 30-minute value 42 of each consumer in the above-mentioned learning data acquisition period from the provided 30-minute value 42 of each consumer (FIG. 4), and for each extracted customer. Based on the 30-minute value 42, the maximum daily power consumption in each season of spring, summer, autumn, and winter (hereinafter referred to as the maximum daily power consumption) and the 30-minute value 42. It has a function of calculating and extracting the maximum value (hereinafter referred to as the 30-minute maximum value), respectively, and storing and managing the calculation result and the extraction result in the power consumption management table 16 described later with respect to FIG. .. In the case of this embodiment, the power consumption management unit 32 sets "spring" as May, "summer" as August, "autumn" as November, and "winter" as February in these months. Calculate and extract the maximum daily power consumption and the maximum 30-minute value, respectively.

The equipment information management unit 33 extracts the installation date of each smart meter from the equipment information provided as described above, and stores and manages the extraction result in the equipment information management table 17 described later with reference to FIG. Has. Further, the elapsed time management unit 34 calculates the number of days elapsed from the installation date of each smart meter based on the equipment information of each smart meter stored in the equipment information management table 17, and the calculation result will be described later with reference to FIG. It has a function of registering and managing the elapsed time management table 18.

The second event count management unit 35 has "1 side power failure", "1 side voltage drop", and "3" detected by each smart meter within a predetermined period based on the event information 40 provided as described above. The number of occurrences of "side voltage drop", "all specific events" and "" 1 side power failure "or" 1 side voltage drop "" is counted for each smart meter and each event type, and the count result is the second event described later. It has a function of storing and managing the number of times management table 19.

The events for which the second event count management unit 35 counts the number of occurrences are the above-mentioned "1 side power failure", "1 side voltage drop", "3 side voltage drop", "all specific events" and "1 side". Of "power outage" or "voltage drop on one side", only the events related to the case where the equipment failure below the low voltage drop line has not occurred yet. Therefore, the second event frequency management unit 35 includes "three-side voltage drop", "one-side power failure", and "one-side power failure" that do not include the stoppage (power failure) of power supply from the electric wire connected to the three-side terminal of the smart meter. The number of occurrences of "1 side voltage drop", "all specific events" and "" 1 side power failure "or" 1 side voltage drop "" is counted respectively.

The prediction model generation unit 36 uses various information stored in the first event count management table 14, the event leveling table 15, the power consumption management table 16, the equipment information management table 17, and the elapsed time management table 18 as learning data. , A prediction model 20 for predicting the occurrence of such equipment defects by machine learning the correlation between the frequency of occurrence of specific events for each type of smart meter and the occurrence of equipment defects below the low-voltage drop line is provided for each type of smart meter. Each has a function to generate.

In practice, as shown in FIG. 8, the prediction model generation unit 36 includes information 50A indicating the number of occurrences of each specific event for each smart meter stored in the first event number management table 14 (FIG. 4), and the event. Information 50B indicating the number of occurrences of events at each level for each smart meter stored in the leveling table 15 (FIG. 4) and consumers corresponding to each smart meter stored in the power consumption management table 16 (FIG. 4). Information 50C indicating the daily maximum power consumption and the maximum 30-minute value in each season, and information 50D indicating the installation date of each smart meter stored in the equipment information management table 17 (Fig. 4) to the consumer. , Information 50E stored in the elapsed time management table 18 (FIG. 4) indicating the elapsed time from the date of installation to the customer for each smart meter is collected for each smart meter using the instrument ID as a key. Generate 50.

Then, the prediction model generation unit 36 generates, for example, a decision tree model as shown in FIG. 9 as a prediction model 20 by using an existing method based on the learning data 50 generated in this way.

The model evaluation unit 37 has a function of evaluating the accuracy of the prediction model 20 generated by the prediction model generation unit 36. In practice, the model evaluation unit 37 has among various data stored in the first event count management table 14, the event leveling table 15, the power consumption management table 16, the equipment information management table 17, and the elapsed time management table 18, respectively. The verification data 51 is input to the prediction model 20 by using various data related to equipment defects (power failure) below the low-voltage drop line that have already been dealt with as verification data 51.

As shown in FIG. 10, the learning data 50 used by the prediction model generation unit 36 to generate the prediction model 20 is long-term data in order to acquire more samples until the equipment fails below the low-pressure drop line. On the other hand, if the verification data 51 used by the model evaluation unit 37 for the accuracy verification of the prediction model 20 is data for a period during which such verification can be performed (hereinafter, this is referred to as a verification period). good. However, in addition to the data of the verification period, various data of a predetermined period (hereinafter, this is referred to as an addition period) are added and used as the verification data 51. This is to enable the prediction of equipment defects below the low voltage drop line that occurred before and after the start of the verification period using the prediction model 20.

In this case, as shown in FIG. 11, the occurrence of a specific event detected by the smart meter corresponding to the occurrence of such a defect on the horizontal axis is the number of days in the past from the date when the equipment defect below the low-voltage drop line occurred. When I made a graph of some examples with the cumulative number of times as the vertical axis, 80% of the specific events detected until the equipment failure below the low voltage drop line finally occurred, the equipment failure occurs. It was confirmed that it occurred 60 days ago. Therefore, in this embodiment, the additional period is set to 60 days, and the verification period is also 60 from the time when the last specific event occurs or the time when there is a track record of dealing with equipment defects below the low-voltage service line. The period up to the day before shall be selected as the verification period.

Then, the model evaluation unit 37 compares the prediction result of the equipment defect below the low-voltage drop line obtained by inputting the above verification data into the prediction model 20 with the equipment defect below the actual low-voltage drop line, thereby performing the prediction model. Evaluate accuracy. In this case, such evaluation is performed by calculating the hit rate and the coverage rate of the occurrence prediction of equipment defects by the prediction model 20.

Here, as shown in FIG. 12A, the set of events in which equipment failure (power failure) of the low-pressure service line actually occurs is "A", and the set of events predicted by the prediction model to cause equipment failure of the low-pressure service line. When is "C", the set "A" and the set "B" which is an overlapping part of the set "C" are a set of events for which the prediction model can correctly predict the equipment failure of the low pressure drop line.

Therefore, the hit rate can be calculated by dividing the number of events belonging to the set "B" by the number of events belonging to the set "C" (B / C), and the coverage rate is calculated by dividing the number of events belonging to the set "C" by the number of events belonging to the set "C". It can be calculated by dividing (B / A) the number of events belonging to "B" by the number of events belonging to the set "A". The relationship between the set "A", the set "B" and the set "C" when the hit rate is 100% is as shown in FIG. 12B, and the set "A" when the coverage rate is 100%. , The relationship between the set "B" and the set "C" is as shown in FIG. 12 (C).

Then, the model evaluation unit 37 displays on the display device 12 (FIG. 1) the hit rate and the coverage rate of the occurrence of equipment defects below the low-voltage drop line using the prediction model 20 calculated in this way.

As a result, the user can confirm the prediction accuracy of the prediction model 20 created at that time based on the display result. Further, when the prediction accuracy is low, the user repeats regenerating the prediction model 20 by giving the learning data 50 for a longer period than the previous time to the prediction model generation unit 36. As a result, the accuracy of the prediction model 20 can be gradually improved. Then, when the accuracy of the prediction model 20 reaches a certain level or higher, the user gives an instruction to the prediction model generation device 1 to predict the occurrence of equipment defects below the low-voltage drop line using the prediction model 20. ..

As a result, the defect occurrence prediction unit 38, which receives such an instruction, receives the data stored in the event leveling table 15, the power consumption management table 16, the equipment information management table 17, and the elapsed time management table 18 at that time. Of the data stored in the second event count management table 19, data that can be a sign of equipment failure below the low-pressure drop line that has not yet been discovered is taken in as actual data 52, and this actual data 52 is used as the prediction model 20. It predicts equipment defects below the low-voltage drop line that will be discovered or will occur in the future. Then, the defect occurrence prediction unit 38 notifies the equipment maintenance staff of the company or the consignment company of the prediction result 53, and instructs the inspection of the equipment below the predicted low-voltage service line.

As shown in FIG. 10, the actual data 52 may be data for a period immediately before (hereinafter, referred to as an analysis period) to the extent that equipment defects below the low voltage drop line can be predicted from the present time. A period in which a predetermined additional period is added in addition to the analysis period is used as the actual data acquisition period, and various data in the actual data acquisition period are used as the actual data 52. This is so that the equipment failure of the low-voltage drop line that occurred before and after the start of the analysis period can be predicted by using the prediction model 20.

On the other hand, the first event number management table 14 holds the number of occurrences of each specific event detected by each smart meter within the latest predetermined period counted by the first event number management unit 30 (FIG. 4). It is a table used for management. As shown in FIG. 13, the first event count management table 14 includes a control number column 14A, an instrument ID column 14B, a one-side voltage drop column 14C, a one-side power failure column 14D, and a three-side voltage drop column 14E. Will be done. In the first event count management table 14, one row corresponds to one smart meter.

A management number (serial number in the present embodiment) assigned to the information in the row in the first event count management table 14 is stored in the control number column 14A, and the information is stored in the instrument ID column 14B. An identifier (instrument ID) unique to the smart meter assigned to the smart meter corresponding to the above is stored.

Further, in the 1-side voltage drop column 14C, the number of occurrences of "1-side voltage drop" detected by the corresponding smart meter within a predetermined period counted by the first event count management unit 30 is stored, and the 1-side voltage drop column 14C is stored. The power failure column 14D stores the number of occurrences of "one-side power failure" detected by the smart meter within the predetermined period, which is counted by the first event number management unit 30. Further, in the 3 side voltage drop column 14E, the occurrence of "3 side voltage drop (including 3 side power failure)" detected by the corresponding smart meter within the predetermined period counted by the first event count management unit 30. The number of times is stored.

Therefore, in the case of the example of FIG. 13, the smart meter with the instrument ID of "instrument A" performs "one-side voltage drop" "11" times, "one-side power failure" "1" times, and "1" times within the predetermined period. It is shown that "3 side voltage drop" was detected "0" times.

The event leveling table 15 is a "one-side power failure", "one-side voltage drop", "three-side voltage drop", "all specific events" and "one-side power failure" leveled by the event leveling unit 31 (FIG. 4). This is a table used to hold and manage the aggregated results of the number of occurrences of events such as "1 side voltage drop" or "1 side voltage drop", "1 side power failure", "1 side voltage drop", and "3 side". It is created for each of "voltage drop", "all specific events" and "" 1-side power failure "or" 1-side voltage drop "", respectively. As shown in FIG. 14, the event leveling table 15 includes a reference number column 15A, an instrument ID column 15B, and a plurality of level columns 15C. In the event leveling table 15, one row corresponds to one smart meter.

The control number (serial number in the present embodiment) assigned to the information of the row in the event leveling table 15 is stored in the reference number column 15A, and the smart corresponding to the information is stored in the instrument ID column 15B. The instrument ID of the smart meter assigned to the meter is stored.

Further, each level column 15C is provided corresponding to each of the five levels 1 to 5, and each level column 15C stores the number of occurrences of the corresponding level counted by the event leveling unit 31. Will be done.

Therefore, in the case of the example of FIG. 14, for the smart meter with the instrument ID of "instrument A", the number of detections of the event leveled to level 1 ("Lv.1") is "10" times and level 2 ("Lv.1"). The number of detected events leveled to .2 ") is leveled to" 4 ", level 3 ("Lv.3 "), level 4 ("Lv.4 "), and level 5 ("Lv.5 "). It is shown that the number of detections of the events was "0".

The electric power consumption management table 16 includes the maximum value of the daily electric power consumption for each season calculated and extracted by the electric power consumption management unit 32 (hereinafter, this is referred to as the daily maximum value). It is a table used to manage the maximum value of the 30-minute value (hereinafter referred to as the maximum value of the 30-minute value). As shown in FIG. 15, the power consumption management table 16 includes a reference number column 16A, an instrument ID column 16B, a seasonal maximum value column 16C, and a 30-minute value maximum value column 16D. In the power consumption management table 16, one row corresponds to one smart meter installed in one consumer.

The reference number column 16A stores the control number (serial number in the present embodiment) assigned to the information in the row in the power consumption management table 16, and the instrument ID column 16B corresponds to the information. The instrument ID of the smart meter assigned to the smart meter is stored.

Further, in the daily maximum value column 16C for each season, the power consumption amount (daily maximum value) of the day when the power consumption amount was the maximum in each corresponding season is stored, and the 30-minute value maximum value column 16D for each season is stored. Stores the maximum 30-minute value (maximum 30-minute value) in each corresponding season.

As described above, in this embodiment, "winter" is set to February, "spring" is set to May, "summer" is set to August, and "autumn" is set to November. In the maximum value column 16C and the 30-minute value maximum value column 16D, the daily maximum value and the 30-minute value maximum value in February are stored, respectively, and the daily maximum value column 16C and the 30-minute value maximum value column corresponding to "spring" are stored. The daily maximum value and the 30-minute maximum value in May are stored in 16D, respectively. In addition, the daily maximum value column 16C and the 30-minute value maximum value column 16D corresponding to "summer" store the daily maximum value and the 30-minute value maximum value in August, respectively, and the daily maximum value column corresponding to "autumn". In the 16C and the 30-minute value maximum value column 16D, the daily maximum value and the 30-minute value maximum value in November are stored, respectively.

Therefore, in the case of the example of FIG. 15, for a consumer who has a smart meter with an instrument ID of "instrument A", the daily maximum value of "winter" is "550", the 30-minute maximum value is "40", and " The daily maximum value for "spring" is "500", the daily maximum value for 30 minutes is "30", the daily maximum value for "summer" is "600", the daily maximum value for 30 minutes is "50", and the daily maximum value for "autumn" is "". It is shown that the maximum value for 30 minutes was "30".

The equipment information management table 17 is a table used to hold and manage the installation date of each smart meter acquired by the equipment information management unit 33 (FIG. 4), and as shown in FIG. 16, the reference number column 17A, It is configured to include an instrument ID column 17B and an instrument installation date column 17C. In the equipment information management table 17, one row corresponds to one smart meter.

The reference number column 17A stores the control number (serial number in the present embodiment) assigned to the information in the row in the power consumption management table 16, and the instrument ID column 17B corresponds to the information. The instrument ID of the smart meter assigned to the smart meter is stored. Further, the date when the corresponding smart meter is attached to the corresponding consumer is stored in the instrument attachment date column 17C.

Therefore, in the case of the example of FIG. 16, it is shown that the smart meter with the instrument ID "Instrument A" is attached to the consumer side corresponding to "2018/3/16".

Further, the elapsed time management table 18 is calculated by the elapsed time management unit 34 (FIG. 4) based on the installation date of each smart meter stored in the equipment information management table 17 to the customer's destination. It is a table used for holding and managing the number of days elapsed since it was attached to the customer, and as shown in FIG. 17, it is configured to include a reference number column 18A, an instrument ID column 18B, and an elapsed time column 18C. Will be done. In the elapsed time management table 18, one row corresponds to one smart meter.

A control number (serial number in the present embodiment) assigned to the information in the row in the elapsed time management table 18 is stored in the reference number column 18A, and the instrument ID column 18B corresponds to the information. The instrument ID of the smart meter assigned to the smart meter is stored. Further, in the elapsed time column 18C, the elapsed time (the number of days in the present embodiment) since the corresponding smart meter is installed at the customer's destination is stored.

Therefore, in the case of the example of FIG. 17, it is shown that the number of days since the smart meter with the instrument ID "instrument A" is installed at the corresponding customer destination is "808" days.

Since the second event count management table 19 has the same configuration as the first event count management table 14 except that the actual data 52 of the actual data acquisition period described above is stored in FIG. The detailed explanation in is omitted.

(2) Operation and effect of the prediction model generation device of the present embodiment In the prediction model generation device 1 of the present embodiment having the above configuration, "one-side power failure", "one-side voltage drop", and "three-side voltage" Specific events corresponding to each event such as "decrease", "all specific events" and "" 1-side power failure "or" 1-side voltage decrease "" are leveled, and the leveling result, the number of occurrences of each specific event, and each Learning data 50 is generated based on the maximum daily power consumption and the maximum 30-minute value for each season of the consumer and the elapsed time since each smart meter was installed on the consumer side, and the generated learning. A prediction model 20 is generated based on the data 50.

In this case, predicting equipment defects below the low-voltage drop line using the leveling results of a specific event is based on the empirical rule that the frequency of failures increases as the life of equipment and devices approaches. This is based on the assumption that the occurrence of equipment defects below the low-pressure drop line will approach as the number of occurrences of events increases, and by using the leveling results of specific events in this way, a more accurate prediction model 20 is generated. It is thought that it can be done.

Further, in this prediction model generation device 1, by using the daily power consumption maximum value and the 30-minute value maximum value of each customer for each season as training data, the power usage of the consumer that generally changes every season is used. A prediction model 20 that takes into account the amount can be generated, and by using equipment information such as the elapsed time since each smart meter is installed on the customer side as learning data 50, the prediction model 20 that also takes into consideration deterioration over time is taken into consideration. Is thought to be able to generate.

Therefore, according to the present prediction model generation device 1, a highly accurate prediction model 20 can be generated, and based on this prediction model 20, the occurrence of equipment defects below the low-voltage drop line can be predicted with high accuracy. By taking advance measures such as parts replacement before the occurrence of such equipment failure, it is possible to increase the cost of maintenance management work of the distribution system due to the emergency commissioned work and reduce the frequency of power outage response. Therefore, according to the present prediction model generation device 1, it is possible to improve the quality of the distribution service and eliminate the busy feeling of responding to a power outage while reducing the cost required for the maintenance and management of the distribution system.

(3) Other Embodiments In the above-described embodiment, the case where the predictive model generation device is configured by one computer device is described, but the present invention is not limited to this, and the present invention is not limited to this, for example, via a network. It may be configured by a plurality of computer devices connected to each other. In this case, the first event number management unit 30, the event leveling unit 31, the power consumption management unit 32, the equipment information management unit 33, the elapsed time management unit 34, the second event number management unit 35, and the prediction model generation. The unit 36, the model evaluation unit 37, and the defect occurrence prediction unit 38 may be distributed and arranged in these a plurality of computer devices.

Further, in the above-described embodiment, as described above with respect to FIG. 8, the information 50A indicating the number of occurrences of each specific event for each smart meter stored in the first event number management table 14 (FIG. 4) and the event. Information 50B indicating the number of occurrences of each level event for each smart meter stored in the leveling table 15 (FIG. 4) and the consumer corresponding to each smart meter stored in the power consumption management table 16 (FIG. 4). Information 50C indicating the daily maximum power consumption and the maximum 30-minute value in each season, and information 50D indicating the installation date of each smart meter stored in the equipment information management table 17 (FIG. 4) to the consumer. , Information 50E stored in the elapsed time management table 18 (FIG. 4) indicating the elapsed time from the date of installation to each consumer is collected for each smart meter using the instrument ID as a key, and the learning data 50. However, the present invention is not limited to this, and a part of these information may be omitted, or other information may be added as training data 50 in addition to or in place of these information. You may.

Further, in the above-described embodiment, the case where various information held by the control cancellation discount system 2, the meter data management system 3, and the various systems 4 is manually provided to the prediction model generation device 1 has been described. The invention is not limited to this, for example, the prediction model generation device 1 is connected to the control stop discount system 2, the meter data management system 3 and various systems 4 via a network, and the prediction model generation device 1 is the control stop discount system 2. , The necessary information may be acquired from the meter data management system 3 and various systems 4 via the network.

The present invention can be applied to a predictive model generator that generates a predictive model for predicting the occurrence of equipment defects below the low voltage drop line.

1 ... predictive model generator, 10 ... CPU, 13 ... analysis tool, 14 ... first event count management table, 15 ... event leveling table, 16 ... power usage management table, 17 ... equipment Information management table, 18 ... Elapsed time management table, 19 ... Second event count management table, 20 ... Prediction model, 30 ... First event count management unit, 31 ... Event leveling department, 32 ... Power usage management department, 33 ... Equipment information management department, 34 ... Elapsed time management department, 35 ... Second event frequency management department, 36 ... Prediction model generation department, 37 ... Model evaluation department, 38 ... ... Defect occurrence prediction unit, 40 ... Event information, 42 ... 30 minutes value, 43 ... Configuration information, 44 ... Equipment information, 50 ... Learning data, 51 ... Verification data, 52 ... Actual data, 53 …… Forecast result.

Claims (10)

In a predictive model generator that generates a predictive model for predicting the occurrence of equipment defects below the low-voltage drop line.
It was detected by the smart meter within a predetermined first period based on the frequency of occurrence of each specific event recognized based on the event information of each specific event detected by the smart meter installed in each consumer. An event leveling unit that levels each specific event to a plurality of levels and counts the number of times the specific event occurs at each specific event for each specific event.
A predictive model generation device including a predictive model generation unit that generates the prediction model based on the number of occurrences of each specific event for each level counted by the event leveling unit. The first period is
The predictive model generation device according to claim 1, wherein the specific event that is a sign of equipment failure below the drop line that occurred at the end of the first period is included in a predetermined ratio or more. The predictive model generation unit
1. 2. The predictive model generator according to 2. Further provided with a power consumption management unit that calculates the maximum value of the 30-minute value for each season and the maximum value of the power consumption for each day based on the 30-minute value of each consumer measured by the smart meter. ,
The predictive model generation unit
Claim 1 is characterized in that the prediction model is generated based on the number of occurrences of each specific event for each level, the maximum value of the 30-minute value for each season, and the maximum value of the daily power consumption. The predictive model generator according to any one of claims 3. The specific event is
The predictive model generation device according to any one of claims 1 to 4, wherein the event is a sign of equipment failure. A predictive model generation method executed by a predictive model generator that generates a predictive model for predicting the occurrence of equipment defects below the low-voltage drop line.
It was detected by the smart meter within a predetermined first period based on the frequency of occurrence of each specific event recognized based on the event information of each specific event detected by the smart meter installed in each consumer. A first step of leveling each specific event to a plurality of levels and counting the number of times the specific event of each level has occurred for each specific event.
A predictive model generation method comprising a second step of generating the predictive model based on the number of occurrences of each of the specific events counted for each level. The first period is
The predictive model generation method according to claim 6, wherein the specific event that is a sign of equipment failure below the drop line that occurred at the end of the first period is included in a predetermined ratio or more. In the second step,
6 or claim 6, wherein the predictive model is generated based on the number of occurrences of each particular event for each level and the elapsed time from the date of attachment of each smart meter to the consumer. 7. The predictive model generation method according to 7. In the first step,
Based on the 30-minute value of each consumer measured by the smart meter, the maximum value of the 30-minute value for each season and the maximum value of the daily power consumption are calculated.
In the second step,
6. The predictive model generation method according to any one of claims 8. The specific event is
The predictive model generation method according to any one of claims 6 to 9, wherein the event is a sign of equipment failure.

Images