JP7739765B2 - Power supply monitoring system and power supply monitoring method - Google Patents

Description

The present invention relates to a power supply monitoring system and a power supply monitoring method.

In recent years, smart meters have been installed in a significant number of consumer premises, including homes, office buildings, factories, and schools, and various systems utilizing the information acquired by smart meters have been proposed.

For example, Patent Document 1 describes an electric power data processing system that processes electric power data transmitted from a smart meter, which generates processed data by adding to the electric power data electric power attribute data relating to the attributes of the electric power amount when the electric power amount constituting the electric power data for a certain period is measured, acquires the user's activity state as activity state data for a specific period arbitrarily set within the certain period corresponding to the generated certain period of processed data, and estimates the user's activity state for the remaining period remaining after subtracting the specific period from the certain period based on a trained model that has been machine-learned to learn the user's activity state corresponding to the processed data using learning data generated from the acquired activity state data and processed data for the specific period corresponding to the specific period during which the activity state data was acquired.

For example, Patent Document 2 describes a distribution line abnormality monitoring system designed to monitor abnormalities in distribution lines from the power supply intake point to the distribution board. The distribution line abnormality monitoring system includes a smart meter for measuring the amount of power supplied from a pole transformer via the intake line, a distribution board with a main breaker connected to the smart meter via the distribution line, and a sensor for measuring the current and voltage flowing through the main breaker, and determines abnormalities in the distribution line based on the difference between the primary measurement value measured by the smart meter and the secondary measurement value measured by the sensor.

Japanese Patent Application Laid-Open No. 2020-28208 JP 2017-17794 A

Currently, power suppliers such as electric power companies do not have a system in place to directly detect abnormalities in the power supply to consumers, and instead respond passively by initiating on-site investigations when consumers report that their electricity is flickering. Therefore, from the perspective of improving the quality of power supply services, power suppliers need to establish a system that enables them to proactively and quickly detect abnormalities occurring on the consumer side and take appropriate action.

The present invention was made in light of this background, and aims to provide a power supply monitoring system and a power supply monitoring method that enable power suppliers to actively and early detect abnormalities occurring on the consumer side.

One of the present inventions for achieving the above-mentioned objectives is a power supply monitoring system configured using an information processing device having a processor and a storage device, communicatively connected to one or more smart meters installed on the consumer side that measure distribution line voltage values, acquiring time series data of the distribution line voltage values measured by the smart meters, storing an anomaly determination model, which is a machine learning model that learns from previously acquired time series data as learning data and outputs a first probability indicating the degree of anomaly in the time series data, inputting newly acquired time series data from the smart meters into the anomaly determination model, acquiring the probability that the time series data contains an anomaly, and outputting information based on the acquired first probability.

For example, the power supply monitoring system compares the first probability with a preset threshold to determine whether or not there is an abnormality in the time series data and outputs the result.

This invention enables early detection of abnormalities occurring on the consumer side based on time-series data of voltage values sent from smart meters. This allows power suppliers to respond quickly and appropriately to abnormalities occurring on the consumer side, resulting in thorough voltage management and improved quality of power supply services.

Another aspect of the present invention is the power supply monitoring system described above, which stores an anomaly determination model, which is a machine learning model that learns the time series data as learning data and outputs the first probability and a second probability, which is the probability that the type of anomaly in the time series data is a predetermined type; by inputting the time series data newly acquired from the smart meter into the anomaly determination model, the first probability and the second probability are acquired, and information based on the acquired first probability and second probability is output.

For example, the power supply monitoring system compares the second probability with a preset threshold to determine the type of abnormality in the time series data and output the result.

According to the present invention, information on the type of abnormality occurring on the consumer side can be obtained based on the time-series data of voltage values sent from the smart meter. This allows the power supplier to identify the type of abnormality occurring on the consumer side and take appropriate action.

Another aspect of the present invention is the power supply monitoring system described above, which stores an anomaly determination model, a machine learning model that learns the time series data as learning data and outputs the first probability, the second probability, and a third probability, which is the probability that the cause of the anomaly in the time series data is a predetermined cause; by inputting the time series data newly acquired from the smart meter into the anomaly determination model, the first probability, the second probability, and the third probability are acquired, and information based on the acquired first probability, second probability, and third probability is output.

For example, the power supply monitoring system compares the third probability with a preset threshold to determine the cause of the abnormality in the time series data and output the result.

According to the present invention, information on the cause of an abnormality occurring on the consumer side can be obtained based on the time-series data of voltage values sent from a smart meter. This allows power suppliers to identify the cause of an abnormality occurring on the consumer side and take appropriate action.

Furthermore, the problems disclosed in this application and the solutions thereto will be made clear in the detailed description and drawings.

This invention enables power suppliers to proactively and early detect abnormalities occurring on the consumer side.

FIG. 1 is a diagram illustrating a schematic configuration of a power supply monitoring system. FIG. 2 is a diagram illustrating main functions of the power supply monitoring device. 1 is a diagram illustrating an example of a hardware configuration of an information processing device used to realize a power supply monitoring device. 1 is an example of a hardware configuration of a smart meter (SM). 10 is an example of smart meter information. 10 is an example of measurement value information. 10 is a graph showing time series changes in measured voltage values under normal conditions. 10 is a graph showing time series changes in measured voltage values when an abnormality occurs. 10 is a graph showing time series changes in measured voltage values when an abnormality occurs. FIG. 10 is a diagram (neural network) illustrating an abnormality determination model. 10 is a flowchart illustrating an abnormality determination model learning process. 10 is an example of a learning data setting screen. 10 is an example of a learning data setting screen. 10 is an example of a learning data setting screen. 10 is a flowchart illustrating an abnormality determination process. 10 is an example of a determination result presentation screen.

One embodiment will be described below with reference to the drawings. In the following description, identical or similar components will be designated by common reference numerals and their description may be omitted.

Figure 1 shows the general configuration of an information processing system (hereinafter referred to as "power supply monitoring system 1") that uses smart meters to monitor the power supply status at consumers, as explained as one embodiment. As shown in the figure, power supply monitoring system 1 includes a power supply monitoring device 100 and a number of smart meters (hereinafter referred to as "SM200") installed at consumers 2 (homes, public facilities, office buildings, various public facilities, factories, infrastructure facilities, etc.) that receive power supply through a power transmission and distribution system.

The power supply monitoring device 100 is installed, for example, in a monitoring station of a power supplier such as an electric power company or in a smart grid control center (hereinafter collectively referred to as the "power supply monitoring center 3"). The power supply monitoring device 100 is configured using an information processing device (computer) and is communicatively connected to the SM 200 installed in each consumer 2 via a communication network 5 by wireless or wired communication. The communication network 5 is configured using a communication infrastructure such as a WAN (Wide Area Network), a concentrator, a 920 Hz band communication network, or a PLC (Power Line Communication).

The SM200 of each consumer 2 transmits various measurement values (voltage, current, active power, reactive power, power factor, temperature, etc.) related to the power supplied from the power transmission and distribution system to the consumer 2 where it is installed to the power supply monitoring device 100 of the power supply monitoring center 3 via the communication network 5. The power supply monitoring device 100 monitors the presence or absence of abnormalities in the power supply at the consumer 2 by analyzing the time series data of measurement values sent from the SM200 at regular intervals (e.g., every 5 minutes, every 30 minutes, etc.) using a machine learning model. Note that in the following description, the term "measurement value" refers to the voltage value (standard voltage 100V (tolerance range 101V±6V), standard voltage 200V (tolerance range 202V±20V)) of the distribution line (single-phase two-wire, single-phase three-wire, three-phase three-wire, etc.) supplying power to the consumer 2.

Figure 2 shows the main functions of the power supply monitoring device 100. As shown in the figure, the power supply monitoring device 100 has the following functions: a memory unit 110, a smart meter information management unit 120, a measurement value acquisition management unit 130, a learning data generation unit 135, a model learning unit 140, an abnormality determination unit 145, and a determination result presentation unit 150.

Of the above functions, the memory unit 110 stores the following information (data): smart meter information 111, measurement value information 112, learning data 113, anomaly determination model 114, and determination results 115.

Of these, the smart meter information 111 includes various information about each SM200 installed in each consumer premises 2.

Measurement value information 112 includes time series data of measurement values acquired from the SM200 of each consumer 2 via the communication network 5.

The learning data 113 is learning data (teaching data) used to train a machine learning model (hereinafter referred to as the "abnormality determination model") used to determine whether or not there is an abnormality in the power supply at the consumer 2 and to identify the type of abnormality.

The abnormality determination model 114 is the entity of the abnormality determination model (represented, for example, by a determinant, mathematical expression, vector, etc., including adjustable parameters). In this embodiment, the abnormality determination model 114 is a deep neural network (DNN), but it may also be realized by other types of models such as gradient boosting (GBDT (Gradient Boosting Decision Tree)), and the type of the abnormality determination model 114 is not necessarily limited.

The judgment result 115 includes information regarding the analysis results of whether or not there is an abnormality in the power supply at the consumer 2, performed using the abnormality judgment model 114.

The measurement value acquisition management unit 130 shown in the same figure receives measurement values sent from the SM200 of each consumer 2 via the communication network 5, and stores time series data based on the received measurement values as measurement value information 112. In this embodiment, the measurement value acquisition management unit 130 acquires measurement values from the SM200 of each consumer 2 at predetermined time intervals (e.g., every 5 minutes, every 30 minutes, etc.) and manages them as time series data. The measurement value acquisition management unit 130 manages both time series data of measurement values newly acquired from the SM200 and time series data of measurement values previously acquired from the SM200 as measurement value information 112.

The learning data generation unit 135 generates learning data to be used for training the anomaly determination model 114. The learning data generation unit 135 presents the user with time-series data of measurement values acquired in the past for a predetermined period (half a day, a full day, a week, etc.; hereinafter referred to as the "time interval") as explanatory variables (features), and accepts from the user the setting of a label (abnormal or normal, type of abnormality if abnormal, etc.) which is the objective variable (correct answer data) for the measurement values. The learning data generation unit 135 then generates data that associates the time-series data of the measurement values with the accepted label, and manages this as learning data 113.

The above-mentioned time intervals are set empirically, for example, based on time-series changes in measurement values when abnormalities occurred in the past. Alternatively, multiple anomaly determination models 114 may be generated for each of multiple time intervals, and the generated multiple anomaly determination models 114 may be used to determine the presence or absence of an anomaly in the power supply at the consumer 2 and the type of anomaly from various perspectives. The above-mentioned time intervals are set to a length that improves the accuracy of anomaly detection, for example, based on past cases.

The model learning unit 140 uses the learning data 113 to learn the anomaly determination model 114. The model learning unit 140, for example, inputs the learning data 113 into the anomaly determination model 114, and then learns the anomaly determination model 114 by adjusting the parameters that define the anomaly determination model 114 using a method such as backpropagation based on the difference between the results output by the anomaly determination model 114 and the input learning data 113.

The abnormality determination unit 145 determines whether there is an abnormality in the power supply at the consumer 2 by inputting the measurement values for the specified period for which the presence or absence of an abnormality is to be determined (for example, the measurement values for the latest specified period newly acquired from the SM 200) into the abnormality determination model 114, and manages the result as the determination result 115.

The judgment result presentation unit 150 generates a screen (judgment result presentation screen 1200, described below) for presenting the contents of the judgment result 115 to the user, and presents (displays) the generated judgment result presentation screen to the user via the user interface.

Figure 3 shows an example of the hardware configuration of an information processing device used to implement the power supply monitoring device 100. The illustrated information processing device 10 includes a processor 11, a main memory device 12, an auxiliary memory device 13, an input device 14, an output device 15, and a communication device 16. Specific examples of information processing devices 10 include personal computers, office computers, various server devices, and general-purpose machines. All or part of the information processing device 10 may be implemented using virtual information processing resources provided using virtualization technology, process space separation technology, or the like, such as a virtual server provided by a cloud system. The power supply monitoring device 100 may also be implemented using multiple information processing devices 10 connected to each other so that they can communicate with each other.

In the same diagram, the processor 11 is configured using, for example, a CPU (Central Processing Unit), MPU (Micro Processing Unit), GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), AI (Artificial Intelligence) chip, etc.

The main memory device 12 is a device that stores programs and data, and is, for example, ROM (Read Only Memory), RAM (Random Access Memory), or non-volatile memory (NVRAM (Non Volatile RAM)).

The auxiliary storage device 13 may be, for example, an SSD (Solid State Drive), a hard disk drive, an optical storage device (CD (Compact Disc), DVD (Digital Versatile Disc), etc.), a storage system, an IC card, a read/write device for recording media such as an SD card or optical recording media, or a storage area of a cloud server. Programs and data can be read into the auxiliary storage device 13 via a recording media reader or communication device 16. Programs and data stored in the auxiliary storage device 13 are read into the main storage device 12 as needed.

The input device 14 is an interface that accepts input from outside, and may be, for example, a keyboard, mouse, touch panel, card reader, pen-input tablet, or voice input device.

The output device 15 is an interface that outputs various information such as processing progress and results. The output device 15 is, for example, a display device (LCD (Liquid Crystal Display), graphics card, etc.) that visualizes the various information described above, a device that converts the various information described above into audio (audio output device (speaker, etc.)), or a device that converts the various information described above into text (printer, etc.). Note that the information processing device 10 may also be configured to input and output information to and from other devices via the communication device 16, for example.

The input device 14 and output device 15 form a user interface that accepts and presents information to the user.

The communication device 16 is a device that realizes communication (wired communication or wireless communication) with other devices via a communication infrastructure such as the communication network 5, and is configured using, for example, a network interface card (NIC), a wireless communication module, a USB module, etc. The communication device 16 communicates with the SM 200 installed in each consumer 2 via the communication network 5. The communication device 16 also communicates with other information processing systems that manage information related to the power transmission and distribution system and the SM 200 installed in each consumer 2, such as a power distribution control system and a customer management system that operate in the power supply monitoring center 3.

The information processing device 10 may be equipped with, for example, an operating system, a file system, a DBMS (DataBase Management System) (relational database, NoSQL, etc.), a KVS (Key-Value Store), etc.

The functions of the power supply monitoring device 100 are realized when the processor 11 of the information processing device 10 reads and executes a program stored in the main memory device 12, or by the functions of the hardware (FPGA, ASIC, AI chip, etc.) that makes up the power supply monitoring device 100. The power supply monitoring device 100 stores the various pieces of information (data) described above, for example, as database tables or files managed by a file system.

Figure 4 shows an example of the hardware configuration of SM200. As shown in the figure, SM200 includes a processor 201, memory 202, a timing device 203, a measurement device 204, an input device 205, an output device 206, a communication device 207, and a power supply device 208.

The processor 201 is configured using a CPU, MPU, power calculation engine (PCE), etc., and realizes the various functions provided by the SM 200 by reading and executing programs stored in the memory 202. The memory 202 is configured using ROM, RAM, non-volatile memory, etc., and stores programs (including firmware) and data. The programs stored in the memory 202 can be updated as needed, for example, by updating firmware, etc.

The timing device 203 is configured using, for example, a temperature-compensated RTC (Real Time Clock) and provides time information such as the current date and time. The time information provided by the timing device 203 is used as information indicating the measurement date and time of measurement information such as voltage measurement values. The timing device 203 may also have a function to obtain date and time information via the communication network 5 using, for example, the Network Time Protocol (NTP).

The measuring device 204 is equipped with various sensors (voltage sensor, current sensor, temperature sensor, etc.), a ΔΣ A/D converter, a gain amplifier, etc., and acquires various measurement values measured at the consumer 2 where the SM 200 is installed and stores them in the memory 202.

The input device 205 and output device 206 provide a user interface that enables workers and others to directly operate the SM200 at the site where the SM200 is installed. The input device 205 is, for example, a keyboard, a touch panel, an operation button, etc. The output device 206 is, for example, a display device such as an LCD, an organic electro-luminescence panel, or an LED (light emitting diode).

The communication device 207 realizes communication with the power supply monitoring device 100 via the communication network 5. The communication device 207 includes, for example, a wireless communication module (such as a 920 MHz band wireless module or a 1:N wireless module) or a wired communication module (such as a PLC (Power Line Communication) module).

The power supply unit 208 supplies drive power to each component of the SM200, and includes, for example, an A/D converter that converts AC power from the power distribution line into DC power, a battery that enables the SM200 to operate independently when the power supply from the power distribution line is interrupted, and a charging circuit for the battery.

Figure 5 shows an example of smart meter information 111. As shown in the figure, smart meter information 111 is made up of multiple records each containing the items of meter ID 1111, consumer ID 1112, and various information 1113. One record of smart meter information 111 corresponds to one smart meter (SM200) installed at consumer 2. Smart meter information 111 is obtained, for example, from a power distribution control system, a customer management system, etc.

Of the above items, meter ID 1111 stores the smart meter identifier (hereinafter referred to as "meter ID"). Consumer ID 1112 stores the consumer identifier (hereinafter referred to as "consumer ID"). Various information 1113 stores various information related to the SM200, such as the type of power supply contract (contracted amperes (10A, 30A, 50A, etc.)) for the consumer 2 in which the SM200 is installed, the section of the distribution line in which the SM200 is installed, identification information for the distribution line (line name, service pole number), and the date and time the SM200 was installed at the consumer 2.

Figure 6 shows an example of measurement value information 112. As shown in the figure, measurement value information 112 is made up of multiple records with items such as meter ID 1121, measurement date and time 1122, voltage value (1) 1123, and voltage value (2) 1124. One record in measurement value information 112 corresponds to a measurement value sent from a certain SM 200 at a certain measurement date and time.

Meter ID 1121 stores the meter ID of the SM200 that measured the measurement value of the record. Measurement date and time 1122 stores the date and time when the measurement value was measured. Voltage value (1) 1123, voltage value (2) 1124, etc. store the measurement values of each measurement point on the distribution line that is the measurement target of the SM200.

Figures 7A to 7C each show graphs of time series data of measurement values (time series data of voltage value (1) 1123 and time series data of voltage value (2) 1124) for a certain time interval in the measurement value information 112. Figure 7A is an example of time series data of voltage values for a time interval during which power supply to consumer 2 is operating normally, and Figures 7B and 7C are both examples of time series data of voltage values for a time interval during which an abnormality is occurring in the power supply to consumer 2.

As shown in FIG. 7A, during time periods when power is being supplied normally, the voltage values (voltage value (1) 1123 and voltage value (2) 1124) fall within the allowable range of 101V ± 6V. On the other hand, as shown in FIGS. 7B and 7C, during time periods when an abnormality is occurring in the power supply, there are periods when voltage value (1) 1123 and voltage value (2) 1124 do not fall within the allowable range of 101V ± 6V. For example, in FIG. 7B, there is a period when the voltage value temporarily falls below 90V. Also, for example, in FIG. 7C, there are periodic periods when the voltage value temporarily falls below 90V. By learning this time series data of voltage values under normal conditions and time series data of voltage values under abnormal conditions, the abnormality determination model 114 determines whether power is being supplied normally and, if an abnormality occurs, identifies the type and cause of the abnormality from newly acquired time series data of voltage values.

Figure 8 shows an example of the anomaly detection model 114 (neural network structure). As shown in the figure, time series data of measurement values over a predetermined time interval, acquired from the measurement value information 112, is input to the input layer 811 of the anomaly detection model 114. The intermediate layer 812 includes one or more hidden layers consisting of one or more nodes containing parameters adjusted by learning. The intermediate layer 812 calculates one or more predicted values (probabilities) for the output layer 813 based on the time series data provided to the input layer 811. In the anomaly detection model 114 of this embodiment, the output layer 813 includes multiple predicted values (probability that an anomaly has occurred in the power supply, probabilities indicating the possibility of each type of anomaly (probability that the occurring anomaly is A, probability that the occurring anomaly is B, probability that the occurring anomaly is caused by α, probability that the occurring anomaly is caused by β, ...)).

In this embodiment, the input layer 811 of the anomaly determination model 114 is provided with time series data of a single measurement value acquired from a certain SM 200 over a predetermined time interval. However, for example, time series data of multiple measurement values acquired from a certain SM 200 over a predetermined time interval at different measurement points on a distribution line may be provided to the input layer 811. By providing the input layer 811 with time series data of multiple measurement values acquired at different measurement points on a distribution line, it is expected that the accuracy of the anomaly determination model 114 will improve.

Figure 9 is a flowchart illustrating the processing performed by the learning data generation unit 135 and model learning unit 140 of the power supply monitoring device 100 when generating learning data and learning the anomaly determination model 114 using the generated learning data (hereinafter referred to as the "abnormality determination model learning processing S900"). The timing for executing the abnormality determination model learning processing S900 is not necessarily limited. However, for example, the power supply monitoring device 100 executes the abnormality determination model learning processing S900 when new learning data 113 is generated or when an execution instruction is received from a user via a user interface. The abnormality determination model learning processing S900 will be described below with reference to this figure.

First, the learning data generation unit 135 of the power supply monitoring device 100 presents the user with the measurement values (explanatory variables) to be used as learning data 113, accepts the user's setting of labels (objective variables) for the measurement values, and generates data that associates the measurement values with the labels as learning data 113 (S911).

Figures 10A to 10C show examples of screens (hereinafter referred to as "learning data setting screen 1000") that the learning data generation unit 135 presents to the user via the user interface when accepting the presentation of the above-mentioned measurement values (explanatory variables) and the setting of labels (target variables). Figure 10A is an example of the learning data setting screen 1000 when the power supply is normal, and Figures 10B and 10C are examples of the learning data setting screen 1000 when the power supply is abnormal. As shown in these screens, the learning data setting screen 1000 has a display field 1010 for the measurement values (explanatory variables) and a field 1020 for setting the labels.

Returning to FIG. 9 , the model learning unit 140 of the power supply monitoring device 100 then learns the anomaly determination model 114 based on the learning data 113 (S912). Note that the model learning unit 140 may, for example, verify the prediction accuracy of the learned anomaly determination model 114. In this case, for example, the learning data 113 is pre-classified into learning data and verification data, and the anomaly determination model 114 is learned using the learning data, and the anomaly determination model 114 is verified using the verification data.

Figure 11 is a flowchart illustrating a process (hereinafter referred to as "abnormality determination process S1100") in which the abnormality determination unit 145 of the power supply monitoring device 100 inputs explanatory variables to be determined (e.g., newly acquired time-series data of measurement values (voltage values) over a specified time interval) into the abnormality determination model 114 to determine objective variables (probability that an abnormality has occurred, probability that the abnormality type is A, probability that the abnormality type is B, ..., probability that the cause of abnormality A is α, probability that the cause of abnormality B is β, ...), and the determination result presentation unit 150 of the power supply monitoring device 100 presents the determined objective variables to the user. The abnormality determination process S1100 is executed at a predetermined timing (periodically, at a scheduled date and time, when an execution instruction is received from the user, etc.). The abnormality determination process S1100 will be described below with reference to this figure.

First, the abnormality determination unit 145 obtains time series data of newly acquired measurement values (voltage values), which will be the explanatory variables of the determination target, from the measurement value information 112 (S1111).

Next, the abnormality determination unit 145 inputs the acquired explanatory variables into the abnormality determination model 114 and obtains the output of the abnormality determination model 114 as the determination result 115 (S1112).

Next, the judgment result presentation unit 150 generates a screen (hereinafter referred to as the "judgment result presentation screen 1200") that displays the judgment result 115, and outputs (displays) the generated judgment result presentation screen 1200 to the output device 15 (display device) (S1113).

Figure 12 shows an example of a judgment result presentation screen 1200. The illustrated judgment result presentation screen 1200 has a display field 1210 for the time period to be judged, and a display field 1220 for the judgment result. The display field 1210 for the time period to be judged displays information indicating the time period of the time-series data of the measurement value to be judged. The display field 1220 for the judgment result displays the contents of the judgment result 115 obtained in S1112 of Figure 11. As shown in the figure, the display field 1220 for the judgment result displays a table consisting of multiple rows with the following items: meter ID 1221, normal/abnormal 1222, abnormality type 1223, and predicted cause 1224. One row in the display field 1220 for the judgment result corresponds to the judgment result of one SM200.

Meter ID 1221 displays the meter ID.

Normal/Abnormal 1222 displays information (normal/abnormal) indicating whether or not an abnormality has occurred in the SM200. For example, if the probability indicating the degree of abnormality in the time series data of measurement values (hereinafter referred to as the "first probability") is below a preset threshold, "Normal" is displayed in Normal/Abnormal 1222. Furthermore, if the first probability exceeds the threshold, "Abnormal" is displayed in Normal/Abnormal 1222.

Anomaly type 1223 displays the predicted type of anomaly if one has occurred. For example, if the probability that the type of anomaly in the time-series data of measured values is a specific type exceeds a preset threshold, anomaly type 1223 displays information indicating the type, such as "Temporary voltage drop" or "Periodic voltage drop."

Cause prediction 1224 displays the predicted cause of an abnormality when one occurs. For example, if the probability that the cause of the abnormality in the time-series data of measurement values is a specific cause exceeds a preset threshold, cause prediction 1224 displays information indicating the cause, such as "poor contact" or "transformer failure." The abnormality determination process S1100 is performed as described above.

As explained in detail above, the power supply monitoring system 1 of this embodiment can detect abnormalities occurring on the consumer side early on, based on the time-series data of voltage values sent from the SM200. This enables the power supplier to respond quickly and appropriately to abnormalities occurring on the consumer side.

Furthermore, with the power supply monitoring system 1 of this embodiment, information regarding the type of abnormality occurring on the consumer side can be obtained based on the time-series data of voltage values sent from the smart meter. This allows the power supplier to identify the type of abnormality occurring on the consumer side and respond appropriately.

Furthermore, with the power supply monitoring system 1 of this embodiment, information can be obtained as to the cause of an abnormality occurring on the consumer side based on the time-series data of voltage values sent from the smart meter. This allows the power supplier to identify the cause of an abnormality occurring on the consumer side and then take appropriate action.

The above describes in detail an embodiment of the present invention, but the above description is provided to facilitate understanding of the present invention and does not limit the present invention. The present invention may be modified or improved without departing from its spirit, and the present invention naturally includes equivalents. For example, the above embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to those that include all of the described configurations. Furthermore, it is possible to add, delete, or replace some of the configurations of the above embodiment with other configurations.

For example, in the above embodiment, the learning data 113 was generated using time-series data of voltage values, which are measurement values, as explanatory variables. However, the learning data 113 may also be generated using, for example, other measurement values (time-series data of current values, time-series data of active power, time-series data of reactive power, time-series data of power factor, time-series data of temperature, etc.) acquired from the measurement values of the SM200 as explanatory variables (features). Furthermore, for example, measurement values acquired from other SM200s (e.g., other SM200s connected to the same transformer) in the vicinity of the SM200 from which the measurement value of interest was acquired, as determined from the smart meter information 111, may also be used as explanatory variables (features). In this way, it is possible to generate learning data 113 that takes into account the conditions surrounding the SM200 from which the measurement value of interest was acquired, and further improvement in the accuracy of the anomaly determination model 114 can be expected.

Furthermore, in the above embodiment, an example has been described in which the anomaly determination model 114 is learned using "supervised learning." However, for example, the anomaly determination model 114 may be generated by inputting a large amount of normal measurement value information 112 into a machine learning model and learning a steady state.

REFERENCE SIGNS LIST 1 Power supply monitoring system, 2 Consumer, 3 Power supply monitoring center, 5 Communication network, 100 Power supply monitoring device, 110 Storage unit, 111 Smart meter information, 112 Measurement value information, 113 Learning data, 114 Abnormality determination model, 115 Determination result, 120 Smart meter information management unit, 130 Measurement value acquisition management unit, 135 Learning data generation unit, 140 Model learning unit, 145 Abnormality determination unit, 150 Determination result presentation unit, 200 SM (smart meter), 1000 Learning data setting screen, 1200 Determination result presentation screen, S900 Abnormality determination model learning process, S1100 Abnormality determination process

Claims (5)

The information processing device is provided on the side of an electric power supplier that supplies electric power to individual consumers through a power transmission and distribution system, and is configured using an information processing device having a processor and a storage device,
A plurality of smart meters are installed on the side of each of the consumers and are communicably connected to measure voltage values of the distribution line,
Acquire time-series data of distribution line voltage values measured at each of the consumers, which is transmitted from each of the smart meters as needed;
storing an anomaly determination model that is a machine learning model that learns the time series data acquired in the past as learning data and outputs a first probability that indicates the degree of anomaly in the time series data;
inputting the time-series data newly acquired from the smart meter into the anomaly determination model to acquire a probability that the time-series data has an anomaly, and outputting information based on the acquired first probability;
storing an anomaly determination model that is a machine learning model that has learned the time series data as learning data and that outputs the first probability and a second probability that the type of anomaly in the time series data is a predetermined type;
inputting the time-series data newly acquired from the smart meter into the anomaly determination model to acquire the first probability and the second probability, and outputting information based on the acquired first probability and second probability;
storing an anomaly determination model that is a machine learning model that has learned the time-series data as learning data and that outputs the first probability, the second probability, and a third probability that is a probability that the cause of each type of anomaly in the time-series data is a predetermined cause;
inputting the time series data newly acquired from the smart meter into the anomaly determination model to acquire the first probability, the second probability, and the third probability, and outputting information based on the acquired first probability, the second probability, and the third probability;
Power supply monitoring system. 2. The power supply monitoring system according to claim 1,
determining whether or not there is an abnormality in the time-series data by comparing the first probability with a preset threshold value, and outputting the result of the comparison;
Power supply monitoring system. 2. The power supply monitoring system according to claim 1 ,
determining whether or not an abnormality exists in the time-series data and the type of the abnormality by comparing the first probability and the second probability with a predetermined threshold value, and outputting the result of the comparison;
Power supply monitoring system. 2. The power supply monitoring system according to claim 1 ,
determining whether or not an abnormality exists in the time-series data, the type of the abnormality, and the cause of the abnormality by comparing the first probability, the second probability, and the third probability with a predetermined threshold value, and outputting the results;
Power supply monitoring system. An information processing device provided on an electric power supplier side that supplies electric power to individual consumers through an electric power transmission and distribution system, the information processing device having a processor and a memory,
acquiring time-series data of the voltage values of the distribution line measured at each of the consumers, the data being sent from each of a plurality of smart meters installed on the side of each of the consumers and measuring the voltage values of the distribution line;
storing an anomaly determination model that is a machine learning model that learns the time series data acquired in the past as learning data and outputs a first probability that indicates the degree of anomaly in the time series data;
inputting the time-series data newly acquired from the smart meter into the anomaly determination model to acquire a probability that the time-series data has an anomaly, and outputting information based on the acquired first probability;
storing an anomaly determination model that is a machine learning model that has learned the time series data as learning data and that outputs the first probability and a second probability that the type of anomaly in the time series data is a predetermined type;
acquiring the first probability and the second probability by inputting the time-series data newly acquired from the smart meter into the abnormality determination model, and outputting information based on the acquired first probability and second probability;
storing an anomaly determination model that is a machine learning model that has learned the time series data as learning data and that outputs the first probability, the second probability, and a third probability that is a probability that the cause of each type of anomaly in the time series data is a predetermined cause;
acquiring the first probability, the second probability, and the third probability by inputting the time-series data newly acquired from the smart meter into the abnormality determination model, and outputting information based on the acquired first probability, the second probability, and the third probability;
A power supply monitoring method for performing the above.