JP2022082294A - Pipe channel abnormality detection system, estimation device, learning model generation device, pipe channel abnormality detector, method for detecting abnormality in pipe channel, method for estimation, and method for generating learning model - Google Patents

Abstract

To easily detect an abnormality in a pipe channel.SOLUTION: An abnormality detection system 100 for a pipe channel includes: an estimation device 10 for estimating weather data with sensor data as an input by using a learning model 80 generated by performing machine learning based on weather data and sensor data from a sensor 50 for detecting the state in a pipe channel as teacher data; and a determination device 20 for comparing the estimated weather data and weather data showing the classification of an actual weather and determining whether an abnormality has happened in the pipe channel.SELECTED DRAWING: Figure 1

Description

The present invention relates to an abnormality detection system for a water channel, an estimation device, a learning model generation device, an abnormality detection device for a water channel, an abnormality detection method for a water channel, an estimation method, and a learning model generation method.

The total length of the basic agricultural pipeline laid by the agricultural and rural development project exceeds 12,000 km, and the total length of its branch canals has reached an enormous length. With the aging of facilities in such agricultural pipelines, sudden damage accidents are on the rise. In particular, many damage accidents have occurred in small-diameter, high-pressure pipe runs for delivering and distributing irrigation water to upland fields.

A crack generated in a pipe body, which is one of the forms of pipeline breakage, grows due to repeated deformation due to water hammer pressure due to water tap operation and water pressure fluctuation due to a pressure reducing valve. The amount of water leakage increases in proportion to the length of the crack that has grown, and eventually the pipe is cut off, leading to a water leakage accident in which a large amount of water is ejected. Therefore, it is important to detect a small amount of water leakage at the initial stage of crack occurrence and prevent a major water leakage accident.

Conventionally, water leakage detection in agricultural pipelines is performed by receiving an abnormal alarm of the flow rate measurement value by TC / TM (telecon telemeter), and after the worker confirms unknown water that leads to water leakage at the site, a sound stick is used. This is done by investigating the exact location of the leak. Further, Patent Documents 1 and 2 describe a technique for detecting water leakage by vibration generated by water leakage. Further, Patent Document 3 describes a technique for detecting water leakage based on a change in flow rate data in a pipeline.

Japanese Unexamined Patent Publication No. 2019-09592 Re-table 2017/199455A. Re-table 2016/181593

However, with the water leakage detection technique using TC / TM or a sound listening stick, it is difficult to detect the water leakage at the initial stage of the occurrence of the water leakage, and the burden on the operator is heavy. Further, there is a demand for a technique capable of detecting water leakage more easily than the techniques described in Patent Documents 1 to 3.

One aspect of the present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a technique for easily detecting an abnormality in a pipe channel.

In order to solve the above problems, the pipe abnormality detection system according to one aspect of the present invention classifies the sensor data acquired from the sensor that detects the state in the pipe and the weather at the time of detection by the sensor. Using a learning model generated by performing machine learning using the represented weather data as teacher data, an estimation device that estimates the weather data using the sensor data as an input, and weather data estimated by the estimation device, are input. It is provided with a determination device for determining the presence or absence of an abnormality in the pipe channel by comparing it with the weather data representing the actual weather classification at the time of acquiring the sensor data.

The estimation device according to one aspect of the present invention performs machine learning using sensor data acquired from a sensor that detects a state in a pipe channel and weather data indicating the classification of the weather at the time of detection by the sensor as teacher data. It is provided with an estimation unit that estimates weather data by using the sensor data as an input using the learning model generated by.

The learning model generator according to one aspect of the present invention performs machine learning using sensor data acquired from a sensor that detects a state in a water channel and weather data indicating the classification of the weather at the time of detection by the sensor as teacher data. By doing so, it is provided with a learning model generation unit that generates a learning model that outputs the weather data by inputting the sensor data.

The abnormality detection device for a conduit according to one aspect of the present invention is a machine using sensor data acquired from a sensor that detects a state in the conduit and weather data indicating the classification of the weather at the time of detection by the sensor as teacher data. Using the learning model generated by learning, the estimation unit that estimates the weather data using the sensor data as input, the weather data estimated by the estimation unit, and the actual operation at the time of acquiring the input sensor data. It is provided with a determination unit for determining the presence or absence of an abnormality in the pipe channel by comparing it with the weather data representing the classification of the weather.

The method for detecting an abnormality in a conduit according to one aspect of the present invention is a machine using sensor data acquired from a sensor that detects a state in the conduit and weather data indicating the classification of the weather at the time of detection by the sensor as teacher data. Using the learning model generated by training, the process of estimating the weather data using the sensor data as input, the weather data estimated in the estimation process, and the actual operation at the time of acquiring the input sensor data. It includes a step of comparing with the weather data representing the classification of the weather and determining the presence or absence of an abnormality in the pipe channel.

In the estimation method according to one aspect of the present invention, machine learning is performed using sensor data acquired from a sensor that detects a state in a water channel and weather data indicating the classification of the weather at the time of detection by the sensor as teacher data. Includes a step of estimating weather data using the sensor data as an input using the learning model generated by.

In the method for generating a learning model of a conduit according to one aspect of the present invention, the sensor data acquired from the sensor that detects the state in the conduit and the weather data indicating the classification of the weather at the time of detection by the sensor are used as teacher data. It includes a step of generating a learning model that outputs the weather data by inputting the sensor data by performing machine learning.

The waterway abnormality detection system according to each aspect of the present invention may be realized by a computer, and in this case, the computer is operated as each part (software element) included in the pipe waterway abnormality detection system. A control program for a waterway abnormality detection system that realizes a waterway abnormality detection system with a computer, and a computer-readable recording medium that records the control program are also included in the scope of the present invention.

According to one aspect of the present invention, it is possible to provide a technique for easily detecting an abnormality in a pipe channel.

It is a block diagram which shows an example of the main part structure of the abnormality detection system of a pipe channel which concerns on one aspect of this invention. It is a figure which shows an example of the teacher data used for the learning model generation in the learning model generation apparatus which concerns on one aspect of this invention. It is a figure which shows an example of the estimation performed by the estimation apparatus which concerns on one aspect of this invention. It is a figure which shows the other example of estimation performed by the estimation apparatus which concerns on one aspect of this invention. It is a flowchart which shows an example of the flow of the learning model generation processing in the learning model generation apparatus which concerns on one aspect of this invention. It is a flowchart which shows an example of the flow of the estimation process in the estimation apparatus which concerns on one aspect of this invention. It is a flowchart which shows an example of the flow of the determination process in the determination apparatus of the abnormality detection system of a pipe channel which concerns on one aspect of this invention.

[Anomaly detection system]
The abnormality detection system for a conduit according to an embodiment of the present invention is a machine using sensor data acquired from a sensor that detects a state in the conduit and weather data indicating the classification of the weather at the time of detection by the sensor as teacher data. Using the learning model generated by training, the estimation device that estimates the weather data using the sensor data as input, the weather data estimated by the estimation device, and the actual sensor data input at the time of acquisition. It is provided with a determination device for determining the presence or absence of an abnormality in the pipe channel by comparing it with weather data representing the classification of the weather.

The present inventors are prone to water leakage due to fatigue failure such as cracking of the pipe body and disconnection of the pipe joint due to deformation in the resin pipe arranged in the agricultural pipeline laid in the agricultural and rural development project. I found that. Leakage in underground agricultural pipelines can lead to road collapse and soil erosion into the surrounding area. Anomaly detection systems for pipelines are applicable to such agricultural pipelines and are particularly suitable for underground buried agricultural pipelines.

As an example of an abnormality in a canal that can be detected by an abnormality detection system for a canal, an abnormality accompanied by water leakage such as cracks in the canal, damage to the canal, breakage of the canal, and disconnection of a joint due to deformation of the canal. Can be mentioned. Further, the abnormality of the pipe canal that can be detected by the abnormality detection system of the pipe canal includes the failure of the device that controls the water supply and drainage through the canal.

FIG. 1 is a block diagram showing an example of a main part configuration of an abnormality detection system 100 for a pipe channel according to an aspect of the present invention. The pipe abnormality detection system 100 includes an estimation device 10 and a determination device 20. The pipe anomaly detection system 100 may further include a learning model generation device 30, a weather data generation device 40, a sensor 50, a storage device 60, and an output device 70.

The sensor 50 detects the state in the pipe channel. The sensor 50 is, for example, a pressure gauge that measures the pressure in the water channel, a flow meter that measures the flow rate in the water channel, or a vibration sensor that detects vibration in the water channel. Further, as the sensor 50, a plurality of sensors may be used in combination, such as a pressure gauge and a flow meter. That is, the sensor 50 detects at least one selected from the group consisting of pressure, flow rate, and vibration in the water channel. The sensor data acquired from the sensor 50 is data representing at least one change within a predetermined time selected from the group consisting of pressure, flow rate, and vibration in the pipe channel. Which sensor is used as the sensor 50 may be selected according to the state of the pipe channel, the installation location, and the like.

Here, the pressure gauge is cheaper than the flow meter and only needs to be attached to the branch portion of the water faucet of the pipe, so that even an existing pipe can be easily attached. Further, if a pressure gauge is used, it is not necessary to consider the attenuation in the pipe channel as in the case of using a vibration sensor, so that data processing is easy and it is easy to detect a change in the state in the conduit. Therefore, the sensor 50 is preferably a pressure gauge. In the present embodiment, the case where the sensor 50 is a pressure gauge will be described as an example, but the present invention is not limited thereto.

It is preferable that a plurality of sensors 50 are provided in a section where an abnormality in the pipe channel is expected to occur, and it is more preferable that the sensors 50 are provided at two locations on the upstream side and the downstream side of the section. That is, the sensor data is data acquired from sensors provided at at least two locations on the upstream side and the downstream side in a predetermined section of the pipe channel. Further, it is more preferable that the sensors 50 are provided at three or more locations including two locations on the upstream side and the downstream side of the above section. This makes it possible to appropriately measure the pressure in the section.

The sensor 50 measures the pressure in the pipe channel, and stores the pressure data (sensor data) including the measured pressure in the storage device 60. This makes it possible to obtain changes in pressure over time. The sensor 50 may be designed to constantly measure the pressure, or may be designed to measure the pressure periodically at predetermined intervals. The sensor 50 may be adapted to transmit pressure data to the storage device 60 wirelessly or by wire. The sensor 50 may be adapted to transmit pressure data to the storage device 60 each time a new pressure is measured, or the pressure data may be stored in the storage device every predetermined time or each time the pressure is measured a predetermined number of times. It may be set to transmit to 60.

The storage device 60 stores programs and data used in the water channel abnormality detection system 100. As an example, the storage device 60 stores the pressure data detected by the sensor 50 in time series at predetermined time intervals. The storage device 60 may have a database for storing pressure data on the cloud or a server.

The storage device 60 may store the pressure data detected by the sensor 50 in chronological order every day (24 hours), and in this case, it is preferable to store the pressure data in association with the date when the pressure is detected. .. Further, the storage device 60 may store the pressure data every half day (12 hours), or may store the pressure data every 6 hours or every 3 hours. The pressure data can be waveform data representing pressure fluctuations within a predetermined time.

The output device 70 outputs the information determined by the determination device 20. The mode of output by the output device 70 is not particularly limited. The output device 70 may be, for example, a display device that displays the information as an image, a printing device that prints the information, or an alarm device that outputs the information as voice.

(Main part configuration of the learning model generator 30)
The learning model generation device 30 performs machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data indicating the classification of the weather at the time of detection by the sensor as teacher data, thereby performing the sensor. A learning model that outputs the weather data is generated by inputting the data.

The learning model generation device 30 includes a control unit 31. The control unit 31 controls each unit of the learning model generation device 30 in an integrated manner, and is realized by a processor and a memory as an example. In this example, the processor accesses the storage (not shown), loads the program (not shown) stored in the storage into the memory, and executes a series of instructions included in the program. As a result, each unit of the control unit 31 is configured. As each of the units, the control unit 31 includes a teacher data acquisition unit 32 and a learning model generation unit 33.

The teacher data acquisition unit 32 acquires teacher data. The teacher data acquisition unit 32 reads out the pressure data stored in the storage device 60 based on the input signal indicating the learning start instruction from the input device (not shown). Reading the pressure data from the storage device 60 can be realized by a known technique such as web scraping.

Further, the teacher data acquisition unit 32 acquires weather data representing the classification of the weather when the pressure included in the read pressure data is detected from the weather data generation device 40. The pressure data read from the storage device 60 is associated with the date on which the pressure was detected. The teacher data acquisition unit 32 generates teacher data in which the date associated with the pressure data and the date associated with the weather data are associated with each other. The teacher data acquisition unit 32 outputs the generated teacher data to the learning model generation unit 33.

When the pressure data is stored in the storage device 60, the weather data of the date associated with the pressure data is acquired from the weather data generation device, and the pressure data and the weather data are stored in association with the date. May be good. As a result, the teacher data acquisition unit 32 can generate the teacher data only by reading the pressure data and the weather data associated with the date from the storage device 60.

FIG. 2 is a diagram showing an example of teacher data 200 used for learning model generation in the learning model generation device 30 according to one aspect of the present invention. As shown in FIG. 2, the teacher data 200 includes date, pressure data, and weather data.

The date represents the date when the pressure data was acquired. The pressure data is data in which the pressure detected by the sensor 50 is stored in the storage device 60 in time series every day, and is waveform data representing fluctuations in the water channel during the day. The weather data is data generated by the weather data generation device 40.

The weather data is data in which the daily weather is classified into predetermined weather classifications, and the weather classifications can be "sunny" and "rainy", but "cloudy", "sunny then cloudy", and "sunny sometimes". It may further include classifications of weather such as "cloudy" and "cloudy temporary rain". The weather data represents the weather on the day the pressure data was acquired.

The teacher data 200 includes one pressure data and one weather data each day. However, when a plurality of sensors 50 are provided in the pipe channel and pressure data from the plurality of sensors 50 are stored in the storage device 60, the teacher data 200 stores the plurality of pressure data and weather data on the same date. Is included.

The learning model generation unit 33 generates a learning model 80 by performing machine learning using the teacher data acquired from the teacher data acquisition unit 32. As an example of the learning model generation unit 33, the learning model 80 is generated by using a known machine learning method such as a neural network, a decision tree, a random forest, or a support vector machine. The learning model 80 generated by the learning model generation device 30 takes pressure data as an input and outputs weather data. The learning model generation unit 33 stores the generated learning model 80 in the storage device 60.

The learning model 80 may be generated by the learning model generation device 30 based on an input signal indicating a learning start instruction from an input device (not shown), or may be periodically performed at predetermined intervals. good. Further, the learning model generation device 30 may be adapted to regenerate the learning model 80 when the pressure data is updated, or may regenerate the learning model 80 every day. The learning model 80 is preferably generated by the learning model generation device 30 for each section of the pipe channel, and may be performed every time the season or the growing stage of the crop changes.

(Main part configuration of estimation device 10)
The estimation device 10 uses a learning model generated by performing machine learning using sensor data acquired from a sensor that detects the state in the pipe channel and weather data representing the classification of the weather at the time of detection by the sensor as teacher data. The weather data is estimated using the sensor data as an input.

The estimation device 10 includes a control unit 11. The control unit 11 controls each unit of the estimation device 10 in an integrated manner, and is realized by a processor and a memory as an example. In this example, the processor accesses the storage (not shown), loads the program (not shown) stored in the storage into the memory, and executes a series of instructions included in the program. As a result, each unit of the control unit 11 is configured. As each of the units, the control unit 11 includes an input data acquisition unit 12 and an estimation unit 13.

The input data acquisition unit 12 acquires pressure data including the pressure detected by the sensor 50. The input data acquisition unit 12 acquires pressure data based on an input signal indicating an estimation start instruction from an input device (not shown). As an example, the input data acquisition unit 12 reads the pressure data from the storage device 60, but may acquire the pressure data directly from the sensor 50. The acquisition of pressure data by the input data acquisition unit 12 can be realized by a known technique such as web scraping. The input data acquisition unit 12 outputs the acquired pressure data to the estimation unit 143.

The estimation unit 13 estimates the weather on the day when the pressure included in the pressure data is detected by using the learning model 80 stored in the storage device 60. The estimation unit 13 reads the learning model 80 from the storage device 60, inputs the pressure data sent from the input data acquisition unit 12 into the learning model 80, and acquires the weather data output from the learning model 80. The estimation unit 13 outputs the acquired weather data to the determination device 20 as an estimation result.

Agricultural pipelines are demand-driven, and pressure fluctuations occur in pipelines when agricultural workers operate (open and close) water taps. That is, the change in the waveform in the pressure data indicates that the faucet was operated. When the weather on the day is fine, the agricultural worker performs water supply work to the field, for example, opening the water tap in the morning and closing the water tap in the afternoon. On the other hand, when the weather on the day is rainy, it is not necessary to supply water to the field, so the agricultural worker does not operate the faucet. Therefore, the pressure data on a sunny day will fluctuate, but the pressure data on a rainy day will not fluctuate.

In this way, the present inventors have described that in the agricultural pipeline, the pressure data in the water pipe fluctuates due to the operation of the faucet of the agricultural worker, and the operation of the faucet by the agricultural worker is the weather of the day. I found that I was influenced by. That is, fluctuations in pressure data in the conduit are affected by the weather on that day. Based on this finding, the present inventors came up with the present invention of estimating weather data by inputting pressure data using a learning model generated by performing machine learning using pressure data and weather data as teacher data. I came to do.

3 and 4 are diagrams showing an example of estimation performed by the estimation device 10 according to one aspect of the present invention. As shown in FIG. 3, the estimation unit 13 inputs the pressure data 300 or the pressure data 301 acquired from the input data acquisition unit 12 into the learning model 80. In the pressure data 300 and the pressure data 301, the pressure fluctuates in the morning and the afternoon. Therefore, when such pressure data is input, the learning model 80 outputs "fine" as the weather data 303. On the other hand, as shown in FIG. 4, the pressure data 400 and the pressure data 401 have no pressure fluctuation or only a slight fluctuation throughout the day. Therefore, when such pressure data is input, the learning model 80 outputs "rain" as the weather data 304.

(Main part configuration of the determination device 20)
The determination device 20 compares the weather data estimated by the estimation device 10 with the weather data representing the classification of the actual weather at the time of acquiring the pressure data input to the learning model, and determines whether or not an abnormality has occurred in the pipe channel. ..

The determination device 20 includes a control unit 21. The control unit 21 controls each unit of the determination device 20 in an integrated manner, and is realized by a processor and a memory as an example. In this example, the processor accesses the storage (not shown), loads the program (not shown) stored in the storage into the memory, and executes a series of instructions included in the program. As a result, each unit of the control unit 21 is configured. As each of the units, the control unit 21 includes a data acquisition unit 22 and a determination unit 23.

The data acquisition unit 22 acquires the estimation result sent from the estimation device 10. The data acquisition unit 22 acquires an estimation result based on an input signal indicating a determination start instruction from an input device (not shown). Then, the data acquisition unit 22 acquires the weather data of the day (the day) associated with the pressure data input to the learning model 80 from the weather data generation device 40. The weather data of the estimation result includes the information of the date associated with the pressure data input to the learning model 80, and the data acquisition unit 22 refers to the information of the date and the weather data of the date. May be acquired from the weather data generation device 40. The data acquisition unit 22 outputs the acquired estimation result weather data and the current day's weather data to the determination unit 23.

The determination unit 23 compares the weather data of the estimation result sent from the data acquisition unit 22 with the weather data of the day, and determines whether or not an abnormality has occurred in the conduit. The determination unit 23 outputs the determination result to the output device 70.

As mentioned above, the fluctuation of the pressure data in the conduit is affected by the weather of the day, so if there is no abnormality in the conduit, the pressure data on a sunny day will fluctuate, but the pressure data on a rainy day. Does not fluctuate. However, any anomalies such as cracks in the conduit will cause fluctuations in the pressure data that are not affected by the weather.

Therefore, if the weather estimated from the pressure data is different from the actual weather of the day, there is a possibility that the pressure data has fluctuations that are not affected by the weather, and that an abnormality has occurred in the conduit. Can be determined. For example, if there is a fluctuation in the pressure data and it is estimated by the estimation device 10 to be "fine", but the actual weather on that day is "rain", the fluctuation in the pressure data is in the waterway. It can be determined that it may be caused by an abnormality.

In this way, the determination unit 23 compares the weather data of the estimation result by the estimation device 10 with the actual weather data acquired from the weather data generation unit. Then, the determination unit 23 determines that "there is an abnormality in the conduit" when the weather data is different, and determines "there is no abnormality in the conduit" when the weather data match.

(Main part configuration of the weather data generation device 40)
The weather data generation device 40 generates weather data in which the weather is classified into a predetermined weather classification based on the acquired text data representing the weather.

The weather data generation device 40 includes a control unit 41. The control unit 41 controls each unit of the weather data generation device 40 in an integrated manner, and is realized by a processor and a memory as an example. In this example, the processor accesses the storage (not shown), loads the program (not shown) stored in the storage into the memory, and executes a series of instructions included in the program. As a result, each unit of the control unit 41 is configured. As each of the units, the control unit 41 includes a text acquisition unit 42 and a classification unit 43.

The text acquisition unit 42 acquires text data representing the weather. The text acquisition unit 42 acquires text data representing the weather based on an input signal indicating an instruction to start acquisition of weather data from the teacher data acquisition unit 32. Further, the text acquisition unit 42 acquires text data representing the weather based on an input signal indicating an instruction to start acquisition of the weather data from the data acquisition unit 22. The text acquisition unit 42 acquires text data representing the weather of the day associated with the pressure data read by the teacher data acquisition unit 32 or the pressure data input to the learning model 80. The text acquisition unit 42 outputs the acquired text data representing the weather to the classification unit 43.

As an example, the text acquisition unit 42 acquires an overview of the weather stored in a database published by an organization that provides meteorological information, such as the Japan Meteorological Agency, as text data representing the weather. The text data representing the weather is data representing the weather of the day associated with the pressure data, but if the pressure data is the pressure data every half day, it may be the text data representing the weather of the half day. Just do it. When the pressure data is pressure data every predetermined time (every 6 hours, every 3 hours), the text data representing the weather may be text data representing the weather at the predetermined time. Further, the text data representing the weather may be text data representing the weather for each region. That is, the text data representing the weather can be text data representing the weather in the area where the section to which the abnormality detection system of the pipe canal is applied exists.

The classification unit 43 generates weather data in which the weather is classified into a predetermined weather classification based on the text data representing the weather sent from the text acquisition unit 42. The classification unit 43 extracts keywords (for example, fine weather, rain, etc.) related to the weather from the acquired text data representing the weather. The extraction of keywords by the classification unit 43 can be realized by a known technique such as text mining.

The classification unit 43 classifies the weather based on the extracted keywords. The classification of the weather can be "sunny" and "rainy", but may further include the classification of the weather such as "cloudy", "sunny then cloudy", "sunny and sometimes cloudy", and "cloudy temporary rain". When the classification of the weather is "fine" and "rain", the classification unit 43 classifies the weather into either "fine" or "rain" based on the extracted keywords and generates the weather data. The classification unit 43 sends the generated weather data to the teacher data acquisition unit 32 or the data acquisition unit 22.

(Flow of learning model generation process)
FIG. 5 is a flowchart showing an example of the flow of the learning model generation process in the learning model generation device 30 according to one aspect of the present invention.

The teacher data acquisition unit 32 reads out the pressure data stored in the storage device 60 based on the input signal indicating the learning start instruction from the input device (not shown) (step S1). The teacher data acquisition unit 32 acquires the weather data of the date from the weather data generation device 40 with reference to the date associated with the read pressure data (step S2). The teacher data acquisition unit 32 associates the pressure data with the weather data having the same date and generates the teacher data (step S3). The teacher data acquisition unit 32 outputs the generated teacher data to the learning model generation unit 33.

The learning model generation unit 33 generates a learning model 80 by performing machine learning using the teacher data acquired from the teacher data acquisition unit 32 (step S4, a step of generating a learning model). The learning model generation unit 33 stores the generated learning model 80 in the storage device 60 (step S5). This completes the learning model generation process (learning model generation method).

(Flow of estimation processing)
FIG. 6 is a flowchart showing an example of the flow of estimation processing in the estimation device 10 according to one aspect of the present invention.

The input data acquisition unit 12 acquires the pressure data of the estimation target stored in the storage device 60 based on the input signal indicating the estimation start instruction from the input device (not shown) (step S11). The input data acquisition unit 12 outputs the acquired pressure data to the estimation unit 13.

The estimation unit 13 inputs the acquired pressure data to the learning model 80 and acquires the output weather data (step S12, estimation step). Then, the estimation unit 13 outputs the acquired weather data as an estimation result to the determination device 20 (step S13). This completes the estimation process (estimation method).

(Flow of judgment process)
FIG. 7 is a flowchart showing an example of the flow of the determination process in the determination device 20 of the pipe channel abnormality detection system 100 according to one aspect of the present invention.

The data acquisition unit 22 acquires the weather data of the estimation result from the estimation device 10 based on the input signal indicating the determination start instruction from the input device (not shown) (step S21). Then, the data acquisition unit 22 acquires the weather data of the day from the weather data generation device 40 (step S22). The data acquisition unit 22 outputs the acquired estimation result and the weather data to the determination unit 23.

The determination unit 23 determines whether or not the weather data of the estimation result sent from the data acquisition unit 22 and the weather data of the current day match (step S23, determination step). When it is determined in step S23 that the weather data of the estimation result and the weather data of the current day match (Yes), the determination unit 23 determines that there is no possibility of an abnormality occurring in the pipe channel (step S24). When it is determined in step S23 that the weather data of the estimation result and the weather data of the day do not match (No), the determination unit 23 determines that there is a possibility of an abnormality occurring in the pipe channel (step S25). Then, the determination unit 23 outputs the determination result to the output device 70 (step S26). This completes the determination process (method for detecting abnormalities in pipe channels).

(effect)
As described above, in the water pipe abnormality detection system 100, the pressure data in the water pipe fluctuates due to the operation of the water tap of the agricultural worker, and the operation of the water tap by the agricultural worker is affected by the weather on the day. It is based on the knowledge found by the present inventors to receive the water supply. If the weather estimated by the learning model and the actual weather of the day match the weather estimated by the learning model by inputting the pressure data measured by the sensor 50, the abnormality detection system 100 of the conduit determines that there is no abnormality in the conduit and is different. In that case, it is determined that there is an abnormality in the pipe channel.

Therefore, the abnormality detection system 100 for the pipe canal can detect the abnormality of the pipe canal only by attaching the sensor 50 to the canal, and there is no need for inspection work at the place where the canal is installed. , The burden on the worker is small. Further, since the pipe waterway abnormality detection system 100 determines the presence or absence of an abnormality in the pipe waterway based on the value measured by the sensor 50, it is possible to make an appropriate judgment unlike the judgment based on the operator's empirical rule. .. Further, since the abnormality detection system 100 for the pipe canal can detect the abnormality by installing the sensor 50 at a position where the abnormality in the canal canal is suspected, it is not necessary to constantly monitor the entire canal.

Further, since the abnormality detection system 100 for the conduit detects an abnormality using the learning model 80 learned by using the pressure data acquired for each predetermined section of the conduit, the abnormality is based on the overall state of the conduit. The accuracy is high compared to the detection of. Further, since the pipe channel abnormality detection system 100 can generate a learning model 80 for each predetermined period or season and use it for detecting the abnormality, it is possible to detect the abnormality according to the predetermined period or the season. And the accuracy is high. Further, the abnormality detection system 100 of the pipe channel can improve the accuracy of abnormality detection only by re-learning the learning model 80 every time the pressure data is accumulated.

Further, if the pressure gauge is used as the sensor 50, the abnormality detection system 100 for the pipe channel is inexpensive and easy to install. Further, if the pressure gauge is used as the sensor 50 in the pipe waterway abnormality detection system 100, it is not necessary to consider the attenuation in the pipe waterway as in the case of using the vibration sensor, so that the data processing is easy and the pipe can be used. It is easy to detect changes in the state of the waterway. Further, if the pressure gauge is used as the sensor 50 in the pipe waterway abnormality detection system 100, the pressure in the pipe waterway changes at the time when a crack occurs in the pipe waterway, so that the pressure in the pipe waterway changes before water leakage occurs from the pipe waterway or. Abnormalities in pipe channels can be detected in the early stages of water leakage.

[Modification example]
The abnormality detection system may be realized as an abnormality detection device including an estimation unit corresponding to the estimation device and a determination unit corresponding to the determination device as a unit. That is, using a learning model generated by performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data. By using the sensor data as an input, the estimation unit that estimates the weather data, the weather data estimated by the estimation device, and the weather data that represents the classification of the actual weather at the time of acquiring the input sensor data are compared. An abnormality detection device for a pipe canal, which is provided with a determination unit for determining the presence or absence of an abnormality in the canal, is also included in the scope of the present invention.

[Example of implementation by software]
The control blocks (particularly the control unit 11, the control unit 21, the control unit 31, and the control unit 41) of the estimation device 10, the determination device 20, the learning model generation device 30, and the weather data generation device 40 are integrated circuits (IC chips). It may be realized by a logic circuit (hardware) formed in the above, or it may be realized by software.

In the latter case, the estimation device 10, the determination device 20, the learning model generation device 30, and the weather data generation device 40 include a computer that executes instructions of a program that is software that realizes each function. The computer includes, for example, one or more processors and a computer-readable recording medium that stores the program. Then, in the computer, the processor reads the program from the recording medium and executes the program, thereby achieving the object of the present invention. As the processor, for example, a CPU (Central Processing Unit) can be used. As the recording medium, a "non-temporary tangible medium", for example, a ROM (Read Only Memory) or the like, a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, or the like can be used. Further, a RAM (Random Access Memory) for expanding the above program may be further provided. Further, the program may be supplied to the computer via any transmission medium (communication network, broadcast wave, etc.) capable of transmitting the program. It should be noted that one aspect of the present invention can also be realized in the form of a data signal embedded in a carrier wave, in which the above program is embodied by electronic transmission.

〔summary〕
The pipe abnormality detection system 100 according to one aspect of the present invention is a sensor data (pressure data) acquired from a sensor 50 that detects a state in the pipe channel, and weather data that represents the classification of the weather at the time of detection by the sensor 50. Using the learning model 80 generated by performing machine learning using It is provided with a determination device 20 for determining the presence or absence of an abnormality in a conduit by comparing it with weather data representing the classification of actual weather at the time of data acquisition.

According to the above configuration, it is possible to detect an abnormality in the water channel only by attaching the sensor 50 to the water channel, and there is no need for inspection work or the like at the place where the water channel is installed. The burden is small. Therefore, the abnormality of the pipe channel can be easily detected.

In the tube anomaly detection system 100 according to one aspect of the present invention, in addition to the above configuration, the sensor data is within at least one predetermined time selected from the group consisting of pressure, flow rate, and vibration in the tube. It is the data showing the change of.

According to the above configuration, it is possible to detect an appropriate abnormality by selecting sensor data according to the state of the pipe channel, the installation location, and the like. Further, if the sensor data is pressure, there is an advantage that the pressure gauge is inexpensive.

In the pipe anomaly detection system 100 according to one aspect of the present invention, in addition to the above configuration, sensor data is collected from sensors 50 provided at at least two locations on the upstream side and the downstream side in a predetermined section of the pipe channel. This is the acquired data. According to the above configuration, the sensor data within a predetermined section can be appropriately acquired.

In addition to the above configuration, the water channel abnormality detection system 100 according to one aspect of the present invention generates weather data in which the weather is classified into a predetermined weather classification based on the acquired text data representing the weather. Further, a data generation device 40 is provided. According to the above configuration, the weather data used for estimation and determination can be acquired.

The estimation device 10 according to one aspect of the present invention performs machine learning using sensor data acquired from the sensor 50 for detecting the state in the pipe channel and weather data representing the classification of the weather at the time of detection by the sensor 50 as teacher data. Using the learning model 80 generated by the operation, the estimation unit 13 for estimating the weather data by inputting the sensor data is provided.

According to the above configuration, it is possible to estimate the weather data used for detecting the abnormality of the pipe channel. Then, it is possible to easily realize the abnormality detection of the pipe channel by using the estimated weather data.

The learning model generator according to one aspect of the present invention uses machine learning as teacher data of sensor data acquired from a sensor 50 that detects a state in a water channel and weather data that represents the classification of the weather at the time of detection by the sensor 50. A learning model generation unit 33 that generates a learning model 80 that outputs weather data by inputting sensor data is provided.

According to the above configuration, it is possible to generate a learning model used for detecting an abnormality in a pipe channel. Then, the generated learning model can be used to easily realize the abnormality detection of the pipe channel.

The tube abnormality detection device according to one aspect of the present invention uses sensor data acquired from the sensor 50 that detects the state in the tube channel and weather data indicating the classification of the weather at the time of detection by the sensor 50 as teacher data. Using the learning model 80 generated by performing machine learning, the estimation unit 13 that estimates the weather data by inputting the sensor data, the weather data estimated by the estimation unit 13, and the actual operation at the time of acquiring the input sensor data. It is provided with a determination unit 23 for determining the presence or absence of an abnormality in the pipe channel by comparing it with the weather data representing the classification of the weather. According to the above configuration, the same effect as that of the pipe waterway abnormality detection system 100 according to one aspect of the present invention can be realized by a single device.

In the method for detecting an abnormality in a conduit according to one aspect of the present invention, the sensor data acquired from the sensor 50 that detects the state in the conduit and the weather data indicating the classification of the weather at the time of detection by the sensor 50 are used as teacher data. Using the learning model 80 generated by performing machine learning, the process of estimating the weather data using the sensor data as input, the weather data estimated in the estimation process, and the actual sensor data at the time of acquisition. It includes a step of comparing with the weather data representing the classification of the weather and determining the presence or absence of an abnormality in the pipe channel. According to the above configuration, the same effect as that of the pipe abnormality detection system 100 according to one aspect of the present invention can be obtained.

In the estimation method according to one aspect of the present invention, machine learning is performed using sensor data acquired from the sensor 50 that detects the state in the pipe channel and weather data indicating the classification of the weather at the time of detection by the sensor 50 as teacher data. The step of estimating the weather data by using the sensor data as an input by using the learning model 80 generated by the above is included. According to the above configuration, the same effect as that of the estimation device 10 according to one aspect of the present invention can be obtained.

In the learning model generation method according to one aspect of the present invention, machine learning is performed using sensor data acquired from the sensor 50 for detecting the state in the pipe channel and weather data representing the classification of the weather at the time of detection by the sensor 50 as teacher data. A step of generating a learning model 80 that outputs weather data by inputting sensor data is included. According to the above configuration, the same effect as that of the learning model generation device 30 according to one aspect of the present invention can be obtained.

The present invention is not limited to the above-described embodiments, and various modifications can be made within the scope of the claims, and the embodiments obtained by appropriately combining the technical means disclosed in the different embodiments. Is also included in the technical scope of the present invention.

10 Estimator 13 Estimator 20 Judgment device 23 Judgment unit 30 Learning model generator 33 Learning model generator 40 Weather data generator 50 Sensor 80 Learning model 100 Pipe channel abnormality detection system

Claims (10)

Using a learning model generated by performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the above An estimation device that estimates the weather data using sensor data as input, and
It is provided with a determination device for determining the presence or absence of an abnormality in the pipe channel by comparing the weather data estimated by the estimation device with the input weather data representing the classification of the actual weather at the time of acquisition of the sensor data. , Anomaly detection system for pipe channels. The abnormality detection system for a water channel according to claim 1, wherein the sensor data is data representing at least one change within a predetermined time selected from the group consisting of pressure, flow rate, and vibration in the water channel. The abnormality detection system for a water channel according to claim 1 or 2, wherein the sensor data is data acquired from sensors provided at at least two locations on the upstream side and the downstream side in a predetermined section of the water channel. The tube according to any one of claims 1 to 3, further comprising a weather data generator that generates the weather data that classifies the weather into a predetermined weather classification based on the acquired text data representing the weather. Anomaly detection system for waterways. Using a learning model generated by performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the above An estimation device equipped with an estimation unit that estimates weather data by inputting sensor data. By performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the sensor data is input and the weather is said. A learning model generator equipped with a learning model generator that generates a learning model that outputs data. Using a learning model generated by performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the above An estimation unit that estimates the weather data using sensor data as input,
A determination unit for determining whether or not an abnormality has occurred in the pipe channel is provided by comparing the weather data estimated by the estimation unit with the input weather data representing the classification of the actual weather at the time of acquisition of the sensor data. , Anomaly detection device for pipe channels. Using a learning model generated by performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the above The process of estimating the weather data using sensor data as input, and
The step includes a step of comparing the weather data estimated in the estimation step with the weather data representing the classification of the actual weather at the time of acquisition of the input sensor data, and determining the presence or absence of an abnormality in the pipe channel. , Anomaly detection method for pipe channels. Using a learning model generated by performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the above An estimation method that includes the process of estimating weather data using sensor data as input. By performing machine learning using the sensor data acquired from the sensor that detects the state in the pipe channel and the weather data representing the classification of the weather at the time of detection by the sensor as teacher data, the sensor data is input and the weather is said. A training model generation method that includes the process of generating a training model that outputs data.

Images