Communication Anomaly Specific Device

Abstract

This invention provides a communication anomaly identification device that can pinpoint the cause and location of communication anomalies in detail. [Solution] The communication anomaly identification device 1 comprises a model generation unit 5 and an estimation unit 6. The model generation unit 5 uses multiple measurement means installed at each of the multiple slave stations provided on the communication line 62 as input data for slave station time-series data measured at time intervals, and generates a model by machine learning from training data which consists of anomaly locations and causes of communication anomalies corresponding to the slave station time-series data as output data. The estimation unit 6 uses newly measured slave station time-series data from the multiple measurement means as estimation input data, and estimates the anomaly locations and causes corresponding to the estimation input data based on the model. The slave station time-series data is associated with slave station location information regarding the location where the slave station is installed.

Description

Technical Field

[0001] The present invention relates to a communication abnormality identification device that identifies the location of a communication abnormality occurring in a communication line provided in a power distribution system and the cause of the communication abnormality.

Background Art

[0002] Conventionally, a power distribution automation system includes a plurality of switches provided on distribution lines constituting a power distribution system, a plurality of slave stations installed for each of the switches, and a master station connected to the slave stations via communication lines. These slave stations remotely monitor and control the switches based on commands from the master station. In addition, communication abnormalities may occur on communication lines, and their symptoms include various cases such as short occurrence duration, recovery after occurrence occurring at night, and repeated occurrence and recovery. Therefore, during on-site investigation after a communication abnormality occurs, if the communication abnormality does not persist, the cause can often not be identified. Furthermore, even if a communication abnormality persists, if the cause is noise, the location where this noise is generated may not be identified. As described above, it is difficult to identify the cause and location of a communication abnormality, and it may be difficult to quickly restore the communication state to normal. Accordingly, in recent years, technologies capable of more quickly discovering abnormal locations in power distribution systems have been developed, and related inventions have already been disclosed.

[0003] Patent Document 1 discloses an invention titled "Power Distribution Line Fault Location Detection System and Method", which relates to a system capable of more quickly detecting a fault location by monitoring switches and the like in a power distribution system. The invention disclosed in Patent Document 1 is a fault detection system in a power distribution system that distributes electricity generated at a power plant to each facility via a substation, a switch, a transformer, and a meter in sequence, and is characterized in that a control unit attached to the substation collects information regarding detection results for the switch and transformer and whether or not there is a response from the meter via a control unit communication means, and includes a control unit determination means that determines whether the switch, transformer, and meter are normal or abnormal based on that information, and a control unit storage means that stores the result. In an invention having these characteristics, by monitoring switches, transformers, and meters in the power distribution system, it is possible to identify the exact location within the faulty section where the fault occurred, and to quickly determine whether or not the cause of the fault lies on the power distribution side. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2008-148413 [Overview of the project] [Problems that the invention aims to solve]

[0005] However, while the invention disclosed in Patent Document 1 can determine whether or not there is a cause of failure on the power distribution side, it cannot grasp the details of the cause of the failure.

[0006] This invention addresses the aforementioned conventional circumstances and aims to provide a communication anomaly identification device that can precisely identify the cause and location of communication anomalies occurring in communication lines. [Means for solving the problem]

[0007] To achieve the above objective, the first invention is a communication anomaly identification device for identifying communication anomalies occurring in communication lines installed in a power distribution system that supplies power from an electric station, comprising a model generation unit and an estimation unit, wherein the model generation unit takes as input data time-series data of slave stations measured at time intervals by a plurality of measuring means installed at each of a plurality of slave stations installed on the communication line, and generates a model by machine learning from training data which consists of anomaly locations and causes of communication anomalies corresponding to the slave station time-series data as output data, and the estimation unit takes as estimation input data newly measured by the plurality of measuring means, and estimates the anomaly locations and causes corresponding to the estimation input data based on the model, wherein the slave station time-series data is associated with slave station location information regarding the location where the slave station is installed.

[0008] In this configuration of the invention, the measuring means provides the slave station with time-series data. The slave station time-series data is measured data of at least one type of physical quantity measured at time intervals. The physical quantity is, for example, voltage, current, and noise of a power distribution line equipped with multiple switches. Furthermore, it is desirable that the substation time-series data include measurement data of normal signals when no communication abnormalities occur, and measurement data of abnormal signals when communication abnormalities occur. Next, the location of the communication anomaly is the location of the slave unit that is causing the communication anomaly among the multiple slave units. The causes of communication anomalies include, for example, a broken communication line, poor contact, and noise generation.

[0009] In the above case, the slave station time-series data is measured data of physical quantities and is associated with slave station location information regarding the location where the slave station is installed. Therefore, the model generation unit learns the correlation between the measured data of physical quantities associated with the slave station location information and the location of the communication anomaly and its cause, and generates a model. Therefore, the estimation unit can estimate the location and cause of the anomaly corresponding to the newly measured physical quantity data based on the model.

[0010] The second invention is characterized in that, in the first invention, the substation location information includes at least one of the distance from the power station to the multiple substations, the addresses of the multiple substations, and the pole numbers of the multiple utility poles on which the multiple substations are each installed, and the substation time-series data includes signal waveform data including a normal signal when no communication abnormality occurs and an abnormal signal when a communication abnormality occurs, as well as status data indicating the occurrence of a communication abnormality and the state of recovery from the communication abnormality.

[0011] In this type of invention, the distance from the substation to multiple substations, the addresses of the multiple substations, and the pole numbers of the multiple utility poles are all factors indicating the location where the substations are installed. Furthermore, the signal waveform data and status data can be associated with the same type of substation location information, or with different types of substation location information. In the latter case, for example, the signal waveform data is substation time-series data relating to the distance from the substation to multiple substations, and the status data is substation time-series data relating to the utility pole numbers.

[0012] In the invention with the above configuration, the signal waveform data and the state data are associated with the same or different types of slave station location information. Therefore, as in the example above, even if the signal waveform data and the state data are in different data formats, the model generation unit learns the correlation between the signal waveform data and the state data and the location of the communication anomaly and its cause, and generates a model. As a result, the types of slave station time-series data that can be used as input data for training data increase, and the estimation unit can estimate the location of the anomaly and its cause based on diverse estimation input data.

[0013] The third invention is characterized in that, in the first or second invention, the input data includes, in addition to the slave station time-series data, the transmission data of the transmission signal transmitted by the power plant and the reply data of the slave station that received the transmission signal and sent back to the power plant with the status of its own station, the output data is the location and cause of the abnormality corresponding to the slave station time-series data, the transmission data and the reply data, and the estimation input data includes, in addition to the slave station time-series data newly measured by the multiple measuring means, the transmission data newly transmitted by the power plant and the reply data newly measured by the multiple measuring means.

[0014] In this configuration, the transmission signal transmitted by the substation is, more precisely, a signal transmitted from a master station installed in the substation to individually monitor multiple slave stations. Therefore, the transmission data and the reply data are associated with slave station location information regarding the location where the slave station is installed. Furthermore, if, for example, the signal level transmitted from the master station is too low for the slave station to receive, the reply data will not be sent back to the master station. In the invention with the above configuration, in addition to the functions of the first or second invention, the estimation unit estimates, for example, that if a particular slave station among a plurality of slave stations does not send reply data, the abnormal location is the master station and the cause of the abnormality is a decrease in the level of the transmission signal transmitted from the master station.

[0015] The fourth invention is characterized in that, in the first or second invention, the communication line is provided with one or more branch boxes in addition to a plurality of substations, and the measuring means connected to the branch box, the branch box is located in a section of the communication line where no substations are provided, at least one of the sections between the power plant and the substations or between the substations, and the branch box is capable of transmitting branch box time-series data measured by the measuring means to the power plant, the branch box time-series data is associated with branch box location information regarding the location where the branch box is located, the input data includes branch box time-series data in addition to substation time-series data and power plant time-series data measured at time intervals by power plant measuring means installed at the power plant, the output data is the abnormal location and cause of occurrence corresponding to the substation time-series data, branch box time-series data and power plant time-series data, and the estimation input data includes branch box time-series data newly measured by the plurality of measuring means in addition to substation time-series data newly measured by the plurality of measuring means and power plant time-series data newly measured by the power plant measuring means.

[0016] In an invention with such a configuration, in addition to the functions of the first or second invention, the model generation unit takes slave station time-series data, junction box time-series data, and power plant time-series data as input data, and generates a model from training data which consists of the location of the communication anomaly and its cause corresponding to the slave station time-series data, junction box time-series data, and power plant time-series data as output data. The estimation unit uses newly measured substation time-series data, newly measured junction box time-series data, and newly measured power plant time-series data as estimation input data, and estimates the location of the communication anomaly and its cause based on the above model. Therefore, even in sections of the communication line where no substations are installed, the location of the communication anomaly and its cause can be estimated.

[0017] The fifth invention is characterized in that, in the first or second invention, the communication line is provided with one or more divisional measurement means installed at the end of a communication line division that divides the communication line, in addition to a plurality of substations, the input data includes substation time series data and one or more divisional time series data measured by one or more divisional measurement means at time intervals, the output data is the substation time series data and the abnormal location and cause of occurrence corresponding to the divisional time series data, the estimation input data includes substation time series data newly measured by a plurality of measurement means and one or more divisional time series data newly measured by one or more divisional measurement means, and the divisional time series data is the substation time series data measured by the measurement means included in the communication line division and the transmission data of the transmission signal transmitted by the power plant.

[0018] In this invention, the substation time-series data includes signal waveform data for the entire communication line, and the segmented time-series data includes signal waveform data for each segment of the communication line. The reason for measuring signal waveform data for each segment of the communication line is that, conventionally, for example, when the cause of an anomaly is noise, the signal waveform data for the entire communication line transmitted to the master station is distorted, making it difficult to pinpoint the location of the anomaly. Therefore, by providing one or more segmented measurement means installed between segments of the communication line, it is possible to distinguish and measure signal waveform data that is not affected by noise from signal waveform data that is affected by noise, thereby improving the accuracy of estimating the location of the anomaly.

[0019] In the invention with the above configuration, in addition to the functions of the first or second invention, the estimation unit uses the substation time-series data newly measured by the plurality of measurement means and the segmented time-series data newly measured by one or more segmented measurement means as estimation input data, and estimates the abnormal location of the communication anomaly and its cause corresponding to this estimation input data based on the above model. Further, the segmented time-series data is slave station time-series data measured by the measurement means included in the communication line segment, and this slave station time-series data is associated with slave station position information. Therefore, by using the segmented time-series data as the estimation input data, the estimation accuracy of the abnormal location is particularly improved. Furthermore, by providing the segmented measurement means, it is considered that the segmented time-series data may differ for each occurrence cause of an abnormality, so the estimation accuracy of the occurrence cause is also improved. Effects of the Invention

[0020] According to the first invention, the estimation unit can estimate the abnormal location and the occurrence cause corresponding to the newly measured slave station time-series data based on the model generated by the model generation unit, so these can be specified in detail. Accordingly, it is possible to greatly reduce the burden on a person when judging the abnormal location of a communication abnormality and its occurrence cause as in the conventional art, and to quickly respond to the abnormal location.

[0021] According to the second invention, in addition to the effects of the first invention, the signal waveform data and the state data are associated with the same type or different types of slave station position information, so the signal waveform data and the state data can be effectively utilized without omission as the input data for training data and the input data for estimation.

[0022] According to the third invention, in addition to the effects of the first or second invention, the estimation unit estimates that the abnormal location is the master station and the occurrence cause is a level drop of the transmission signal transmitted from the master station, so it is possible to grasp whether the cause of the abnormality is on the electric substation side or on the power distribution system side.

[0023] According to the fourth invention, in addition to the effects of the first or second invention, even in sections of the communication line where no substations are installed, the location of the communication anomaly and its cause can be estimated, allowing for a more detailed identification of the anomaly. Therefore, the fourth invention can also contribute to the discovery of potential defects. Furthermore, since the input data for the training data consists of substation time-series data, junction box time-series data, and power plant time-series data, the anomaly can be identified with high accuracy.

[0024] According to the fifth invention, in addition to the effects of the first or second invention, the accuracy of estimating the location of abnormalities and the cause of communication abnormalities is improved, making it possible to pinpoint abnormal locations and the like that were previously difficult to identify. [Brief explanation of the drawing]

[0025] [Figure 1] This is a diagram showing the configuration of a communication anomaly detection device according to an embodiment. [Figure 2] This is a process diagram of the model generation process performed by the communication anomaly detection device according to the embodiment. [Figure 3] This is a process diagram of the communication anomaly identification method performed by the communication anomaly identification device according to the embodiment. [Figure 4] This graph illustrates the signal waveform data included in the time-series data of the substation. [Figure 5] This graph illustrates the signal waveform data included in the time-series data of the substation. [Figure 6] This table explains the state data included in the time-series data of the substations. [Figure 7] This is a table to explain the location information of substations. [Figure 8] This table explains the signal waveform data associated with utility pole numbers. [Figure 9] This is the location of the communication anomaly and its cause, as estimated by the estimation unit. [Figure 10] This is a diagram showing the configuration of the power distribution system targeted by the communication anomaly detection device according to the second modified example of the embodiment. [Figure 11]This is a diagram showing the configuration of the power distribution system targeted by the communication anomaly detection device according to the third modified example of the embodiment. [Modes for carrying out the invention] [Examples]

[0026] A request response support device according to an embodiment of the present invention will be described in detail with reference to Figures 1 to 11. Figure 1 is a configuration diagram of a communication anomaly identification device according to an embodiment. As shown in Figure 1, the communication anomaly identification device 1 according to the embodiment is a device that identifies communication anomalies occurring in a communication line 62 installed in a power distribution system 60 that supplies power from an electric station. Here, an electric station refers to a power plant or a substation. Furthermore, the communication anomaly detection device 1 is specifically a computer and comprises a communication unit 2, a control unit 3, a storage unit 7, and an output unit 12. In addition, the communication anomaly detection device 1 can communicate with a server 51 via a communication channel 52a. This server 51 can communicate with a power distribution control device 50 equipped with a master station 50a via a communication channel 52b, and stores slave station time-series data transmitted from slave stations 63 (see Figures 4 and 5) to the master station 50a via a communication line 62. The slave stations are installed along the communication line 62 after the signal distributor 61 connected to the power distribution control device 50. Furthermore, the time-series data of the substation is associated with substation location information regarding the location where substation 63 is installed.

[0027] The communication unit 2 is a means for connecting the communication anomaly identification device 1 and the server 51 in a communication-enabled manner. The control unit 3 is a central processing unit that controls the operation of the communication anomaly identification device 1, and comprises a time-series data conversion unit 4, a model generation unit 5, and an estimation unit 6. The memory unit 7 comprises a location information memory unit 8, a training data memory unit 9, a model memory unit 10, and an estimated data memory unit 11. The following describes the various components of the control unit 3.

[0028] The time-series data conversion unit 4 converts one of the time-series data of multiple types of substations, each associated with different types of substation location information (for example, the distance from a power plant to multiple substations and the utility pole number), into multiple types of substation time-series data associated with the same type of substation location information (for example, the utility pole number) (hereinafter referred to as substation conversion data), when the input data of the training data stored in the training data storage unit 9 is multiple types of substation time-series data associated with different types of substation location information (for example, the utility pole number). This conversion is performed by the time-series data conversion unit 4 by referring to the input data stored in the training data storage unit 9 and the substation location table (see Figure 7) stored in the location information storage unit 8. However, the time-series data conversion unit 4 is omitted if there is only one type of slave station time-series data (for example, only data on the voltage of power lines) or if multiple types of slave station time-series data are all associated with the same type of slave station location information. In addition, the time-series data conversion unit 4 may perform the same conversion on the estimation input data stored in the estimation data storage unit 11 as described above.

[0029] The model generation unit 5 uses multiple measurement means installed at each of the multiple slave stations 63 located on the communication line 62 as input data, and generates a model using machine learning from training data, which consists of output data representing the location of the communication anomaly and the cause of the communication anomaly, corresponding to the slave station time series data. Note that the measurement means are built into the slave station 63 and are therefore not shown in the diagram. The input data, which is the time-series data of the substation, includes the converted substation data and the time-series data of the substation that has not been converted by the time-series data conversion unit 4.

[0030] The estimation unit 6 uses the newly measured substation time-series data from the measurement means as estimation input data, and estimates the abnormal location and cause of occurrence corresponding to the estimation input data based on the model generated by the model generation unit 5.

[0031] Next, we will explain the configuration of the memory unit 7 and the output unit 12. The location information storage unit 8 stores location information for the locations where the slave stations 63 are installed. Specifically, this location information is stored as a slave station location table in which the distance from the power distribution control device 50 (more precisely, the signal distributor 61) installed in the power station to the multiple slave stations 63, the addresses of the multiple slave stations 63, and the pole numbers of the multiple utility poles on which the multiple slave stations 63 are installed are arranged so that each indicates the same slave station location.

[0032] The training data storage unit 9 stores input data and output data for training data used by the model generation unit 5 to generate a model. The input data is slave station time-series data measured at time intervals by multiple measurement means installed at each of the multiple slave stations 63, and the output data is the location of the communication anomaly and the cause of the communication anomaly, corresponding to this slave station time-series data. The model storage unit 10 stores the model generated by the model generation unit 5. The estimation data storage unit 11 stores the estimation input data input to the estimation unit 6, the location of the communication anomaly corresponding to this estimation input data, and the cause of its occurrence. The output unit 12 is a monitor that displays the location of the communication anomaly and its cause, as estimated by the estimation unit 6.

[0033] Next, the model generation method performed by the communication anomaly identification device will be explained using Figure 2. Figure 2 is a process diagram of the model generation process performed by the communication anomaly identification device according to the embodiment. As shown in Figure 2, the model generation method 20 comprises the time series data conversion step S21 and the model generation step S22. The time-series data conversion process in S21 is a process in which the time-series data conversion unit 4, as described above, refers to the slave station time-series data stored in the training data storage unit 9 and the slave station location table (see Figure 7) and converts multiple types of slave station time-series data associated with different types of slave station location information into slave station conversion data. The slave station conversion data is stored in the training data storage unit 9 together with the slave station time-series data. This process is omitted if there is only one type of slave station time-series data, etc.

[0034] The model generation process in S22 is a process in which the model generation unit 5 generates a model from the training data stored in the training data storage unit 9 using machine learning. The generated model is stored in the model storage unit 10. This completes the model generation method 20.

[0035] Furthermore, the method for identifying communication anomalies performed by the communication anomaly identification device will be explained using Figure 3. Figure 3 is a process diagram of the communication anomaly identification method performed by the communication anomaly identification device according to the embodiment. As shown in Figure 3, the communication anomaly identification method 30 comprises the estimation input data acquisition step S31, the time-series data conversion step S32, the estimation step S33, and the display step S34. Each step will be described below. The estimation input data acquisition step in S31 is a step in which the estimation unit 6 acquires estimation input data from the server 51 via the communication channel 52a and the communication unit 2. The acquired estimation input data is stored in the estimation data storage unit 11.

[0036] The time-series data conversion step in S32 is a process in which the time-series data conversion unit 4 refers to the estimation input data stored in the estimation data storage unit 11 and the slave station location table (see Figure 7) stored in the location information storage unit 8, and converts multiple types of estimation input data associated with different types of slave station location information into multiple types of estimation input data associated with the same type of slave station location information. The converted estimation input data is stored in the estimation data storage unit 11 together with the estimation input data before conversion. Similarly, this step is omitted when there is only one type of estimation input data, etc.

[0037] The estimation process in S33 is a process in which the estimation unit 6 estimates the location of the communication anomaly and its cause, corresponding to the estimation input data converted by the time-series data conversion unit 4 or the estimation input data before conversion. The estimated location of the anomaly and its cause are stored in the estimation data storage unit 11. The display step in S34 is a step in which the output unit 12 displays the estimated location of the communication anomaly and its cause. This completes the communication anomaly identification method 30.

[0038] Next, the time-series data of the substation will be explained using Figures 4 to 6. Figure 4 is a graph illustrating the signal waveform data included in the time-series data of the substation. As shown in Figure 4, the signal waveform data represents the signal level measured for the voltage of the power distribution line over a distance D from the signal distributor 61 to each substation 63, and represents the signal level at a specific point in time in the substation time-series data. The magnitude of the signal level is a relative value with the signal level at the signal distributor 61 set to 1. Signal levels S1 to S5 are theoretical signal levels and represent normal signals when no communication abnormalities occur. Of these, signal level S1 is the signal level from the signal distributor 61 to the branch connection 64a, and signal level S2 is the signal level from the branch connection 64a to the termination resistor 65a. Furthermore, signal level S3 is the signal level from branch connection 64a to branch connection 64b, and signal level S4 is the signal level from branch connection 64b to termination resistor 65b. In addition, signal level S5 is the signal level from branch connection 64b to termination resistor 65c.

[0039] In contrast, signal levels S6 to S8 are abnormal signals that occur when a communication abnormality occurs. Of these, signal level S6 is the signal level from slave station 63a to termination resistor 65b when a communication abnormality occurs due to poor contact at slave station 63a, and signal level S7 is the signal level from branch connection 64b to termination resistor 65c when the same communication abnormality occurs. At this time, the occurrence of communication abnormalities and recovery to normal occurs repeatedly in the section from slave station 63a to termination resistor 65b and in the slave stations arranged from slave station 63a to termination resistor 65c. Furthermore, signal level S8 is the signal level from the branch connection 64b to the termination resistor 65b when a communication abnormality occurs at slave station 63a due to a break in the communication line. At this time, the slave stations downstream from slave station 63a do not return to normal. These signal levels S6 to S8 are below the threshold Th, which is the level at which the power distribution control device 50 can receive signals from each substation 63, and therefore are not received by the power distribution control device 50.

[0040] Next, Figure 5 is a graph illustrating the signal waveform data included in the slave station time-series data. Figure 5 shows signal waveform data when a communication anomaly occurs due to noise N. This communication anomaly is caused by the influx of noise N into slave station 63b, resulting in repeated occurrences of abnormal waveforms and subsequent recovery to normal. Furthermore, because the level of noise N fluctuates, the range of slave stations 63 where the communication anomaly occurs also fluctuates, and this range may extend to slave stations 63c and 63d, for example. As a result, although the signal level exceeds the threshold Th, the overlapping noise N distorts the signal waveform, making it impossible to detect the signal level.

[0041] Furthermore, we will explain the state data included in the slave station time series data. Figure 6 is a table for explaining the state data included in the slave station time series data. As shown in Figure 6, the substation time-series data includes status data indicating the occurrence of communication anomalies and the recovery from those anomalies. This status data represents the occurrence of communication anomalies and recovery from those anomalies for each utility pole number in chronological order. Thus, the status data is associated with the utility pole number among the slave station location information. On the other hand, as shown in Figures 4 and 5, the signal waveform data is associated with the distance D from the signal distributor 61 to each slave station 63 among the slave station location information. Furthermore, the time-series data of the training data input and the newly measured time-series data of the training data input are obtained by continuously measuring signal levels at time intervals, as shown in Figures 4 to 6.

[0042] Furthermore, the contents of the substation location information will be explained using Figure 7. Figure 7 is a table for explaining the substation location information. As can be seen from Figures 4 to 6, the signal waveform data and status data are associated with different types of slave station location information, so it is necessary to convert them into slave station conversion data that is associated with the same type of slave station location information. As mentioned above, this conversion is performed by the time-series data conversion unit 4 by referring to the slave station location table shown in Figure 7.

[0043] As shown in Figure 7, the distance D (m) from the signal distributor 61 to each substation 63, the substation address, and the utility pole number, all listed on the same line, are indicators representing the location of the same substation 63. Therefore, the time-series data conversion unit 4 can convert the distance D of the signal waveform data shown in Figures 4 and 5 into a utility pole number. Thus, this signal waveform data and the status data shown in Figure 6 are associated with a common utility pole number.

[0044] Next, we will explain the signal waveform data associated with the utility pole number using Figure 8. Figure 8 is a table for explaining the signal waveform data associated with the utility pole number. As shown in Figure 8, the signal waveform data is time-series data for each utility pole number. This signal waveform data is output by the time-series data conversion unit 4 (see Figure 1) in the time-series data conversion process of S21 (see Figure 2) and the time-series data conversion process of S32 (see Figure 3). As a result, the measured values ​​from the measuring means installed in each substation 63 are arranged for each utility pole number. Since the measuring means are set to measure the voltage, current, noise, etc. of the distribution line, signal waveform data exists for each of these physical quantities.

[0045] Furthermore, the estimation results estimated by the estimation unit will be explained using Figure 9. Figure 9 shows the location of the communication anomaly and its cause estimated by the estimation unit. As shown in Figure 9, the estimation unit 6 (see Figure 1) estimates and outputs the utility pole number where the communication anomaly occurred, as well as the cause of the communication anomaly, for the date and time the anomaly occurred. On the distribution system 60 side, the causes are a broken communication line, poor contact, and noise. A decrease in transmission level is a cause on the distribution control device 50 side. This decrease in transmission level is output when the input data for training data and the input data for estimation include reply data that sends the status of the slave station back to the power station, and will be explained in the first modified example of the embodiment.

[0046] As explained above, the communication anomaly identification device 1 allows the estimation unit 6 to estimate the location of the communication anomaly and its cause, corresponding to the newly measured slave station time-series data, based on the model generated by the model generation unit 5, thus enabling detailed identification of these. Furthermore, since time-series data for each of several types of physical quantities exists for each of the substations, it is usually difficult and time-consuming for a person to determine the location and cause of a communication anomaly from this signal waveform data and status data. However, with the communication anomaly identification device 1, the estimation unit 6 estimates the location and cause of the communication anomaly, which significantly reduces the burden compared to conventional methods and enables a rapid response to the anomaly.

[0047] Furthermore, with the communication anomaly identification device 1, there is no need to go on an investigation to confirm the location and cause of the anomaly on a day when no communication anomaly has occurred, as was done in the past. This also helps to eliminate unnecessary work. In addition, the communication anomaly identification device 1 allows the time-series data conversion unit 4 to fully and effectively utilize signal waveform data and status data associated with the same type of slave station location information as training data input data and estimation input data. Therefore, based on the diverse estimation input data, the estimation unit 6 can estimate the location of the communication anomaly and its cause, enabling accurate identification of the anomaly location and other related information.

[0048] Next, a communication anomaly identification device 1A according to the first modified example will be described. The configuration of the communication anomaly identification device 1A is the same as that of the communication anomaly identification device 1. In the communication anomaly identification device 1A, the input data for the training data includes signal waveform data, slave station time-series data including status data, transmission data of the transmission signal sent by the power station, and reply data. This reply data is sent back to the power distribution control device 50 by a slave station 63 that receives the transmission signal, indicating its own status. The transmission signal is a signal that the power distribution control device 50 installed in the power station sends periodically at time intervals to individually monitor the multiple slave stations 63. Therefore, the transmitted data and the reply data are time-series data associated with the slave station location information (e.g., utility pole number) regarding the location where the slave station 63 is installed. In addition, if the level of the transmitted signal from the power distribution control device 50 is too low for the slave station 63 to receive, the reply data is not sent back to the power distribution control device 50.

[0049] Furthermore, the output data of the training data includes the time-series data of the slave station, the transmitted data, and the reply data, as well as the location of the communication anomaly and its cause. Furthermore, the input data for estimation includes, in addition to the substation time-series data newly measured by multiple measurement devices, transmission data newly transmitted by the power plant and reply data newly measured by multiple measurement devices.

[0050] In the communication anomaly identification device 1A, the model generation unit 5 takes the slave station time-series data, transmission data, and reply data as input data, and generates a model from training data which consists of the location of the communication anomaly and its cause corresponding to the slave station time-series data, transmission data, and reply data as output data. The estimation unit 6 uses the newly measured time-series data of the substations, transmitted data newly sent by the power plant, and the newly measured reply data as input data for estimation, in addition to the substation time-series data newly measured by the multiple measurement means. Based on the model generated by the model generation unit 5, it estimates the location of the communication anomaly and its cause corresponding to this input data.

[0051] Therefore, the estimation unit 6 estimates that if a specific slave station 63 among the multiple slave stations 63 does not send reply data, the abnormal location is the power distribution control device 50 and the cause is a decrease in the level of the transmission signal transmitted from the power distribution control device 50 (see the bottom row of Figure 9). Thus, it is possible to determine whether the cause of the abnormality lies on the power distribution control device 50 side or on the power distribution system 60 side. The operation and effects of the other communication abnormality identification device 1A are the same as those of the communication abnormality identification device 1.

[0052] Furthermore, a communication anomaly detection device 1B according to a second modified example of the embodiment will be explained using Figure 10. Figure 10 is a diagram of the power distribution system targeted by the communication anomaly detection device according to the second modified example of the embodiment. As shown in Figure 10, the power distribution system 60A targeted by the communication anomaly detection device 1B is equipped with multiple substations 63, one or more branch boxes 66, and measuring means 67 connected to these branch boxes 66. The rest of the configuration of the communication anomaly detection device 1B is the same as that of the communication anomaly detection device 1. The junction box 66 and the measuring means 67 are located in a section of the communication line 62 where no substations 63 are provided, and the measuring means 67 can transmit the time-series data of the junction box measured to the power distribution control device 50. The section where no substations 63 are provided is between the power distribution control device 50 and the substation 63 closest to the power distribution control device 50. However, the junction box 66 and the measuring means 67 may be provided in at least one of the following locations: between the power distribution control device 50 and the nearest substation 63, or between the substations 63 themselves. The master station 50a of the power distribution control device 50 is equipped with a substation measurement means 50b that measures substation time-series data at time intervals. This substation time-series data is time-series data of signal waveform data, noise, etc., exchanged between the master station 50a and the substation 63.

[0053] Furthermore, the branch box time-series data, like the slave station time-series data, is obtained by the measurement means 67 by measuring the voltage of the distribution line, etc., at time intervals, and is associated with branch box location information regarding the location where the branch boxes 66 are placed. This branch box location information can be, for example, the distance from the signal distributor 61 to each branch box 66. In this case, the utility pole number, which is the slave station location information of the status data, is converted by the time-series data conversion unit 4 to the distance from the signal distributor 61 to each slave station 63.

[0054] In the communication anomaly identification device 1B, the input data for the training data includes, in addition to the slave station time-series data, the junction box time-series data and the power plant time-series data. The output data for the training data is the location of the communication anomaly and its cause, corresponding to the slave station time-series data, the junction box time-series data, and the power plant time-series data. Furthermore, the input data for estimation includes, in addition to the slave station time-series data newly measured by the multiple measurement means 67, the junction box time-series data newly measured by the multiple measurement means 67 and the power plant time-series data newly measured by the power plant measurement means 50b.

[0055] Therefore, in the communication anomaly identification device 1B, the model generation unit 5 takes the slave station time-series data, the junction box time-series data, and the power plant time-series data as input data, and generates a model from training data which is the location of the communication anomaly and its cause, corresponding to the slave station time-series data, the junction box time-series data, and the power plant time-series data. The estimation unit 6 uses, in addition to the substation time-series data newly measured by multiple measurement means, the reply data newly measured by one or more measurement means 67 and the power station time-series data newly measured by the power station measurement means 50b as estimation input data, and estimates the location of the communication anomaly and its cause corresponding to this estimation input data based on the model generated by the model generation unit 5. Therefore, the communication anomaly identification device 1B can estimate the location and cause of communication anomalies even in sections of the communication line 62 where no substations 63 are installed, allowing for more detailed identification of the anomaly. Thus, the communication anomaly identification device 1B can also contribute to the discovery of potential faulty areas. Furthermore, since the input data for the training data consists of substation time-series data, junction box time-series data, and power plant time-series data, it can accurately pinpoint the anomaly. The other functions and effects of the communication anomaly identification device 1B are the same as those of the communication anomaly identification device 1.

[0056] A third modified example of the embodiment, specifically the communication anomaly detection device 1C, will be described using Figure 11. Figure 11 is a diagram of the power distribution system targeted by the communication anomaly detection device according to the third modified example of the embodiment. As shown in Figure 11, the power distribution system 60B targeted by the communication anomaly identification device 1C is equipped with multiple substations 63, as well as section measurement means 68a to 68e. The section measurement means 68a to 68e are installed at the ends of the communication line sections 62A to 62D, which are sections of the communication line 62, and measure one or more section time-series data measured at time intervals. Specifically, the segmentation measurement means 68a is provided adjacent to the downstream side of the signal distributor 61. The segmentation measurement means 68b to 68d are provided adjacent to the upstream side of the termination resistors 65a to 65c. Furthermore, the segmentation measurement means 68e is provided in a location between the signal distributor 61 and the branch connection section 64a where no substation 63 is installed. Specifically, three segmentation measuring means 68a, 68e, and 68b are provided between the signal distributor 61 and the termination resistor 65a, and three segmentation measuring means 68a, 68e, and 68c are provided between the signal distributor 61 and the termination resistor 65b. Furthermore, three segmentation measuring means 68a, 68e, and 68d are provided between the signal distributor 61 and the termination resistor 65c. The configuration of the other communication anomaly detection device 1C is the same as that of the communication anomaly detection device 1.

[0057] The segmented time-series data measured by segmented measurement means 68b to 68d consists of substation time-series data measured by multiple measurement means (not shown) included in each of the communication line segments 62A to 62D, and transmission data of transmission signals transmitted by the power distribution control device 50 installed in the power station to each of the substations 63 included in the communication line segments 62A to 62D. This transmission data is time-series data associated with substation location information regarding the location where the substation 63 is installed. In communication line sections 62A to 62D, if, for example, the location of the noise anomaly is at communication line section 62B, which is far from communication line section 62A, the time-series data of the slave station measured at communication line section 62A will be less affected by the noise occurring at communication line section 62B. Therefore, the signal waveforms of the time-series data of the slave stations at communication line sections 62A and 62B will be different and distinguishable.

[0058] Therefore, in the communication anomaly identification device 1C, the model generation unit 5 takes the slave station time-series data and multiple segmented time-series data measured by the segmented measurement means 68a to 68e as input data, and generates a model from this slave station time-series data and training data which are output data representing the location of the communication anomaly and its cause, corresponding to the segmented time-series data. The estimation unit 6 uses, in addition to the substation time-series data newly measured by the multiple measurement means, multiple segmented time-series data newly measured by the segmented measurement means 68a to 68e as estimation input data, and estimates the location of the communication anomaly and its cause corresponding to this estimation input data based on the model generated by the model generation unit 5.

[0059] Therefore, the communication anomaly identification device 1C can estimate the location of the anomaly and its cause based on the time-series data of the substations measured by multiple measurement means on the communication line 62, multiple measurement means included in each of the communication line sections 62A to 62D, the transmission data of the transmission signal transmitted by the power distribution control device 50, and the location of the communication anomaly. This allows for a more detailed identification of the anomaly location and its cause. In particular, the greater the distance between the abnormal location occurring in any of the communication line sections 62A to 62D and the section measurement means 68a to 68e, the less the time-series data of each slave station measured by the section measurement means 68a to 68e will be affected by the abnormality occurring at the abnormal location, and accurate slave station time-series data can be obtained. Therefore, the abnormal location can be estimated with high accuracy from the differences in the signal waveform data of each slave station time-series data. The operation and effects of the communication abnormality identification device 1C other than this are the same as those of the communication abnormality identification device 1.

[0060] The communication anomaly identification devices 1, 1A, 1B, and 1C according to the present invention are not limited to those shown in the embodiments. For example, the slave station time-series data may consist only of signal waveform data. The slave station location information may include at least one of the distance D from the power distribution control device 50 to the plurality of slave stations 63, the addresses of the plurality of slave stations 63, and the pole numbers of the plurality of utility poles on which the plurality of slave stations 63 are installed. Furthermore, the junction box location information may include at least one of the distance from the substation to the junction box, the address of the junction box, and the pole numbers of the utility poles on which the junction boxes are installed. In addition, in the communication anomaly identification device 1C, one of the three sectional measurement means 68a, 68e, and 68b between the signal distributor 61 and the termination resistor 65a may be omitted, or one of the sectional measurement means 68e or 68a, 68b may be omitted. Furthermore, new sectional measurement means may be added in addition to the sectional measurement means 68a to 68e. [Industrial applicability]

[0061] The present invention can be used as a communication anomaly identification device to identify the location and cause of communication anomalies occurring in communication lines installed in a power distribution system. [Explanation of symbols]

[0062] 1, 1A, 1B, 1C…Communication Anomaly Identification Device 2…Communication Unit 3…Control Unit 4…Time Series Data Conversion Unit 5…Model Generation Unit 6…Estimation Unit 7…Storage Unit 8…Location Information Storage Unit 9…Training Data Storage Unit 10…Model Storage Unit 11…Estimation Data Storage Unit 12…Output Unit 20…Model Generation Method 30…Communication Anomaly Identification Method 50…Power Distribution Control Device 50a…Master Station 50b…Electricity Station Measurement Means 51…Server 52a, 52b…Communication Path 60, 60A, 60B…Power Distribution System 61…Signal Distributor 62…Communication Line 62A~62D…Communication Line Classification 63, 63a~63e…Slave Station 64a, 64b…Branch Connection Unit 65a~65c…Termination Resistor 66…Branch Box 67…Measurement Means 68a~68e…Classification Measurement Means

Claims

1. A communication anomaly identification device for identifying communication anomalies occurring in communication lines installed in a power distribution system that supplies power from an electric station, It comprises a model generation unit and an estimation unit. The model generation unit uses time-series data from multiple substations, each equipped with multiple measurement means installed on the communication line, as input data, and generates a model using machine learning from training data, which consists of the location of the communication anomaly and the cause of the communication anomaly corresponding to the substation time-series data as output data. The estimation unit uses the time-series data of the substations newly measured by the multiple measurement means as estimation input data, and estimates the abnormal location and the cause of occurrence corresponding to the estimation input data based on the model. A communication anomaly identification device characterized in that the slave station time-series data is associated with slave station location information regarding the location where the slave station is installed.

2. The substation location information includes at least one of the distance from the substation to the plurality of substations, the addresses of the plurality of substations, and the pole numbers of the plurality of utility poles on which each of the plurality of substations is installed. The communication anomaly identification device according to claim 1, characterized in that the slave station time-series data includes signal waveform data including a normal signal when no communication anomaly occurs and an abnormal signal when a communication anomaly occurs, as well as status data indicating the occurrence of the communication anomaly and the state of recovery from the communication anomaly.

3. The input data includes, in addition to the slave station time-series data, transmission data of the transmission signal transmitted by the power station, and reply data of the slave station that received the transmission signal and sent back to the power station with its own status. The output data corresponds to the time-series data of the substation, the transmitted data, and the reply data, and includes the location of the anomaly and the cause of the anomaly. The communication anomaly identification device according to claim 1 or 2, characterized in that the estimation input data includes, in addition to the slave station time-series data newly measured by the plurality of measurement means, the transmission data newly transmitted by the power plant and the reply data newly measured by the plurality of measurement means.

4. The communication line is provided with a plurality of substations, one or more branch boxes, and the measuring means connected to these branch boxes. The branch box is located in at least one section of the communication line between the substation and the substation, or between the substations themselves, where no substations are provided, and is capable of transmitting the branch box time-series data measured by the measuring means to the substation. The aforementioned branch box time-series data is associated with branch box location information regarding the location where the branch box is placed. The input data includes, in addition to the substation time-series data, the branch box time-series data and the substation time-series data measured at time intervals by the substation measurement means installed in the substation. The output data corresponds to the time-series data of the substation, the time-series data of the junction box, and the time-series data of the power plant, and represents the location of the anomaly and the cause of the anomaly. The communication anomaly identification device according to claim 1 or 2, characterized in that the estimation input data includes, in addition to the substation time-series data newly measured by the plurality of measurement means, the branch box time-series data newly measured by the plurality of measurement means and the power station time-series data newly measured by the power station measurement means.

5. In addition to the multiple substations, the communication line is provided with one or more division measuring means installed at the end of each division of the communication line. The input data includes, in addition to the substation time series data, one or more segmented time series data measured by one or more segmented measurement means at intervals of time. The output data consists of the time-series data of the substation and the location and cause of the anomaly, corresponding to the segmented time-series data. The estimation input data includes, in addition to the substation time series data newly measured by the multiple measurement means, one or more segmented time series data newly measured by one or more segmented measurement means. The communication anomaly identification device according to claim 1 or 2, characterized in that the segmented time-series data is the time-series data of the substation measured by the measuring means included in the communication line segment and the transmission data of the transmission signal transmitted by the power station.

Images