Pipe deterioration prediction system, pipe deterioration prediction method, and program

Abstract

To improve the accuracy of predicting deterioration of piping components. [Solution] A pipe deterioration prediction system is configured comprising: a conversion unit that converts vibration acceleration detected by a sensor, which is provided to detect vibrations corresponding to vibrations occurring in a pipe member to be predicted in a pipe, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency; and a prediction unit that predicts the deterioration status of the pipe member to be predicted based on the vibration evaluation data.

Description

Technical Field

[0001] The present invention relates to a piping deterioration prediction system, a piping deterioration prediction method, and a program.

Background Art

[0002] There is known a technique for estimating the deterioration of a resin pipe member by hitting the surface of the resin pipe member with a hammer equipped with an acceleration sensor and measuring the acceleration detected by the acceleration sensor over time (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the technique described in Patent Document 1, the prediction of the deterioration of the pipe member is not based on an event generated in the pipe member itself, but on the repulsive force of the hammer that hits the pipe member, so the prediction accuracy may not be sufficient.

[0005] In consideration of the above problems, an object of the present invention is to improve the accuracy of predicting the deterioration of a pipe member.

Means for Solving the Problems

[0006] One aspect of the present invention for solving the above problems is a piping deterioration prediction system including: a conversion unit that converts vibration acceleration output from a sensor provided to detect vibration acceleration corresponding to vibration generated in a piping member to be predicted in a pipe into vibration evaluation data as a time waveform, a frequency spectrum, or a natural frequency; and a prediction unit that predicts the deterioration status of the piping member to be predicted based on the vibration evaluation data.

[0007] One aspect of the present invention is a pipe deterioration prediction method in a pipe deterioration prediction system, comprising: a conversion step in which a conversion unit converts vibration acceleration output from a sensor, which is provided to detect vibrations corresponding to vibrations occurring in a pipe member to be predicted in a pipe, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency; and a prediction step in which a prediction unit predicts the deterioration status of the pipe member to be predicted based on the vibration evaluation data.

[0008] One aspect of the present invention is a program for causing a computer in a pipe deterioration prediction system to function as a conversion unit that converts vibration acceleration output from a sensor, which is provided to detect vibrations corresponding to vibrations occurring in a pipe member to be predicted in the pipe, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency, and a prediction unit that predicts the deterioration status of the pipe member to be predicted based on the vibration evaluation data. [Effects of the Invention]

[0009] According to the present invention, the accuracy of predicting deterioration of piping components can be improved. [Brief explanation of the drawing]

[0010] [Figure 1] This figure shows an example of the overall configuration of the pipe deterioration prediction system in the first embodiment. [Figure 2] This figure shows an example of a training dataset in the first embodiment. [Figure 3] This figure shows an example of a procedure corresponding to the prediction of the deterioration status of piping components performed in the piping deterioration prediction system of the first embodiment. [Figure 4] This figure shows an example of a machine learning model according to the second embodiment. [Figure 5] This figure shows an example of a procedure corresponding to the prediction of the deterioration status of piping components performed in the piping deterioration prediction system of the second embodiment. [Modes for carrying out the invention]

[0011] <First Embodiment> Figure 1 shows an example of the overall configuration of the pipe deterioration prediction system in this embodiment. The pipe deterioration prediction system of this embodiment predicts the deterioration status of target pipe components in actual pipes 10 that are actually installed in a building, for example. The pipe deterioration prediction system of this embodiment may be used, for example, by a business operator who has been requested to perform an inspection by a building manager or owner. It may also be used by the manager of a building where the pipes to be deteriorated are located.

[0012] The pipe deterioration prediction system shown in Figure 1 comprises a sensor 100 and a pipe deterioration prediction device 200. Sensor 100 is, for example, an acceleration sensor that detects acceleration (vibration acceleration) corresponding to vibration. In this embodiment, sensor 100 is provided so as to be able to detect vibrations occurring in the piping members that are the target of deterioration prediction among the piping members that make up the actual piping 10. Specifically, sensor 100 may be provided so as to be attached to the target piping member in the actual piping 10.

[0013] In this embodiment, in order to predict the deterioration status of the piping member to be predicted, the worker strikes the piping member to be predicted with a hammer 20 to generate vibrations in the piping member to be predicted. The sensor 100 detects the vibration acceleration corresponding to the vibration generated in the piping member to be predicted.

[0014] The pipe deterioration prediction device 200 is a device that predicts the deterioration status of a target pipe component based on the vibration acceleration detected by the sensor 100. The pipe deterioration prediction device 200 may be configured with hardware such as a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), and storage devices such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). The pipe deterioration prediction device 200 may also be equipped with a GPU (Graphics Processing Unit) as hardware. The functions of the pipe deterioration prediction device 200 shown in the figure are realized by the CPU and GPU in the pipe deterioration prediction device 200 executing a program.

[0015] The pipe deterioration prediction device 200 comprises a data interface unit 201, a user interface unit 202, a control unit 203, and a storage unit 204. The data interface unit 201 handles data exchange. In this embodiment, the data interface unit 201 receives vibration acceleration data output from the sensor 100.

[0016] The user interface unit 202 is equipped with controls and input devices to accept user operations. The user interface unit 202 is also equipped with a display device to display images to the user. Furthermore, the user interface unit 202 is equipped with a speaker or similar device to output sound to the user. The user interface unit 202 may also be equipped with indicators or lamps to notify the user by illuminating these indicators or lamps.

[0017] The control unit 203 performs various controls in the pipe deterioration prediction device 200. The control unit 203 comprises a conversion unit 231 and a prediction unit 232.

[0018] The conversion unit 231 converts the vibration acceleration detected by the sensor 100 into a frequency spectrum (an example of vibration evaluation data) by performing frequency analysis such as FFT (Fast Fourier Transform). That is, the vibration acceleration detected by the sensor 100 is converted into data in the frequency domain by the conversion unit 231. When the vibration acceleration detected by the sensor 100 remains converted into a frequency spectrum, measurement variations may occur in a specific frequency range. Therefore, the conversion unit 231 may perform band limitation so as to exclude the frequency range in which measurement variations occur from the frequency spectrum converted from the vibration acceleration detected by the sensor 100. In addition, there are individual differences among operators in the hammering (hammering) of the piping member to be predicted using the hammer 20, and variations corresponding to the individual differences in hammering occur in a specific frequency range in the frequency spectrum converted from the vibration acceleration detected by the sensor 100. Therefore, the conversion unit 231 may perform band limitation so as to exclude the frequency range in which variations corresponding to the individual differences in hammering occur from the frequency spectrum converted from the vibration acceleration detected by the sensor 100. If one frequency range includes the other frequency range between the frequency range in which measurement variations occur and the frequency range in which variations corresponding to the individual differences in hammering occur, band limitation may be performed using the one frequency range. In this case, as a result of performing band limitation on the one frequency range, the other frequency range will also be band-limited.

[0019] The prediction unit 232 predicts the deterioration state of the piping member to be predicted based on the frequency spectrum obtained by the conversion unit 231. The prediction unit 232 predicts the deterioration state using the machine learning model stored in the machine learning model storage unit 241.

[0020] The machine learning model stored in the machine learning model storage unit 241 may be created as follows. In creating a machine learning model, the creator prepares multiple sample piping components of the same type as the target material, each degraded under different degradation conditions. These sample piping components may have a predetermined length, for example, several tens of centimeters. The degradation conditions may include, for example, the UV irradiation time (an example of a condition value) applied to the sample piping component. Each of these prepared sample piping components will have a different degradation mode. The degradation mode corresponds to the degree of degradation, exposure time, and other degradation conditions.

[0021] As described above, the same sensor as sensor 100 is attached to the deteriorated sample piping member. The creator taps each sample piping member and inputs the vibration acceleration detected by the sensor in response to the tapping into a learning model creation device (not shown). The piping deterioration prediction device 200 may also function as a learning model creation device. Furthermore, the optimal striking position for obtaining good vibration acceleration in the sample piping member is determined by the position where the sensor is attached. Therefore, it is preferable for the creator to strike the sample piping member at the striking position that yields good vibration acceleration. The learning model creation device converts each input vibration acceleration into a frequency spectrum. In this way, multiple frequency spectra corresponding to different degradation modes of sample piping members are obtained. Each frequency spectrum exhibits characteristics specific to the degradation mode of the corresponding sample piping member. Furthermore, the learning model creation device may perform bandwidth limiting on the frequency spectrum used to create the machine learning model, similar to the frequency spectrum used to predict the degradation status, to exclude frequency ranges where measurement variability occurs. In addition, the learning model creation device may perform bandwidth limiting to exclude frequency ranges where variability occurs due to individual differences in hammering.

[0022] The creator creates a training dataset DS by associating labels indicating the corresponding degradation modes for each of the multiple frequency spectra corresponding to the different degradation modes described above. Specifically, as shown in Figure 2, the training dataset DS is formed by including training data for each of the first to nth degradation modes. From the first degradation mode to the nth degradation mode, the UV irradiation time as a degradation-inducing condition is longer.

[0023] Each of the first to nth degradation mode training data is formed by associating the frequency spectrum obtained for the first to nth degradation mode with the first to nth labels indicating that it is the first to nth degradation mode. Furthermore, through augmentation, multiple degradation mode training data sets may be created for a single degradation mode by varying the frequency spectrum.

[0024] The creator inputs the training dataset DS, which consists of the 1st to Nth degradation mode training data created in this manner, into the machine learning model in the learning model creation device. In the learning model creation device, the machine learning model 2411 (Figure 2) is trained using the input training dataset DS. The machine learning model 2411, having undergone this training, outputs an estimated degradation status in response to the frequency spectrum input.

[0025] Furthermore, the machine learning model of this embodiment may also estimate the degradation status by interpolating to the first to nth degradation modes.

[0026] The machine learning model created in this way was found to produce more accurate prediction results compared to predicting the deterioration status of piping components based on the rebound force of a hammer detected when the piping components are struck with a hammer, as demonstrated in the validation experiments.

[0027] Returning to the explanation in Figure 1. The output unit 233 outputs degradation status information indicating the degradation status as predicted by the prediction unit 232. The output unit 233 may output the degradation status information to display on a display device in the user interface unit 202, or it may output the degradation status information as sound from the sound output unit of the user interface unit 202, or it may light up indicators in the user interface unit 202 in a predetermined pattern according to the content of the degradation status information. In addition, the output unit 233 may output the degradation status information to other devices connected, for example, via a network.

[0028] The memory unit 204 stores various information related to the pipe deterioration prediction device 200. The memory unit 204 includes a machine learning model memory unit 241. The created machine learning model is transferred from the machine learning model creation device to the pipe deterioration prediction device 200. The pipe deterioration prediction device 200 stores the transferred machine learning model in the machine learning model storage unit 241. Furthermore, the pipe deterioration prediction device 200 may be configured to also function as a learning model creation device.

[0029] Referring to the flowchart in Figure 3, an example of the procedure corresponding to the prediction of the deterioration status of piping components performed in the piping deterioration prediction system of this embodiment will be described. Step S100: The worker attaches the sensor 100 to the part of the pipe component to be predicted in the actual piping 10. The sensor 100 is connected to the pipe deterioration prediction device 200 so that the detected vibration acceleration can be output to the pipe deterioration prediction device 200. The worker strikes the pipe component to be predicted with a hammer.

[0030] Step S102: The sensor detects the vibration acceleration corresponding to the vibration generated in the pipe member to be predicted by the tapping in step S100.

[0031] Step S104: The vibration acceleration detected by the sensor 100 in step S102 is input to the data interface unit 201 of the pipe deterioration prediction device 200. In the pipe deterioration prediction device 200, the conversion unit 231 acquires the input vibration acceleration. The conversion unit 231 converts the acquired vibration acceleration into a frequency spectrum.

[0032] Step S106: The conversion unit 231 performs bandwidth limiting on the frequency spectrum obtained in step S104 to remove variations such as measurement variations and individual differences in tapping.

[0033] Step S108: Next, the prediction unit 232 calls a machine learning model stored in the machine learning model storage unit 241. The prediction unit 232 inputs the frequency spectrum, which has been bandwidth-limited in step S106, to the called machine learning model.

[0034] Step S110: The machine learning model outputs the degradation mode estimated in response to the frequency spectrum input from step S108.

[0035] Step S112: The prediction unit 232 predicts the deterioration status based on the deterioration mode output by the machine learning model in step S110. The deterioration status predicted by the prediction unit 232 may be, for example, the exposure time, degree of deterioration, remaining lifespan, etc., for the piping component to be predicted.

[0036] Step S114: The output unit 233 generates degradation status information indicating the degradation status predicted in step S112, and outputs the generated degradation status information.

[0037] <Second Embodiment> In buildings, the actual piping 10 may be constructed with a mixture of multiple different types of piping components. Here, the type of piping component may correspond to specifications such as the material, inner diameter, and thickness of the piping component. Thus, different types of piping components result in different frequency spectra. The first embodiment described above uses a single machine learning model to predict the deterioration status for a single type of piping component. Therefore, if other types of piping components are targeted for prediction, it may not be possible to predict the deterioration status with sufficient accuracy. Therefore, the pipe deterioration prediction system of this embodiment, with the configuration described below, is capable of predicting the deterioration status with high accuracy according to the type of pipe component to be predicted.

[0038] Figure 4 shows an example of a machine learning model stored in the machine learning model storage unit 241 of the pipe deterioration prediction device 200 of this embodiment. As shown in the figure, the machine learning model storage unit 241 stores multiple machine learning models A, B, C, etc., which are created to correspond to different types of pipe members A, B, C, etc.

[0039] Figure 5 is a flowchart showing an example of the procedure for predicting the deterioration status of piping components in the piping deterioration prediction system of this embodiment. The processing in steps S200 to S206 is the same as in steps S100 to S106 in Figure 3.

[0040] Step S208: In the pipe deterioration prediction device 200, the prediction unit 232 determines which type of pipe member the frequency spectrum obtained after bandwidth limiting in step S206 corresponds to. In other words, the prediction unit 232 determines the type of pipe member to be predicted. The prediction unit 232 may use a machine learning model to estimate the type of pipe member when determining the type. Such a machine learning model may be created by training it to output the estimated type of pipe member in response to the input frequency spectrum. The machine learning model for estimating the type of pipe member may be stored in the memory unit 204.

[0041] Step S210: The prediction unit 232 retrieves a machine learning model from among the machine learning models stored in the machine learning model storage unit 241 that is associated with the type determined in step S208.

[0042] Step S212: The prediction unit 232 inputs the frequency spectrum, which has been bandwidth-limited in step S206, into the machine learning model called in step S210.

[0043] The processing in steps S214 to S218 is the same as in steps S110 to S114 in Figure 3.

[0044] <Variation> The following describes some variations of the above embodiments. [First variation] In each of the above embodiments, when vibration acceleration is output from the sensor 100 due to the tapping of the piping member to be predicted, the piping deterioration prediction device 200 immediately predicts the deterioration status. As a modified example, the pipe deterioration prediction device 200 may store the vibration acceleration detected by the sensor 100 in the storage unit 204 by performing a tapping motion on each pipe component to be predicted at a site such as a building where an actual pipe 10 containing one or more pipe components to be predicted is installed. Alternatively, the frequency spectrum converted from the vibration acceleration detected by the sensor 100 may be stored in the storage unit 204. After storing the vibration acceleration or frequency spectrum in the storage unit 204 as described above, the pipe deterioration prediction device 200 may, for example, in response to an operation by a user, use the vibration acceleration or frequency spectrum stored in the storage unit 204 to predict the deterioration status of each pipe component to be predicted.

[0045] [Second variation] The functions of the pipe deterioration prediction device 200 in each of the above embodiments may be realized by a plurality of devices, each assigned a predetermined function, cooperating via communication.

[0046] [Third variation] The pipe deterioration prediction device 200 may be configured as a server or cloud server located on a network. In this case, the vibration acceleration output from the sensor 100 in response to the tapping of the pipe component to be predicted may be temporarily received by the worker's terminal, and then transmitted from the worker's terminal via the network to the server or cloud server acting as the pipe deterioration prediction device 200. In this case, the pipe deterioration prediction device 200 may transmit deterioration status information, for example, to the worker's terminal, indicating the predicted deterioration status of the pipe component.

[0047] [Fourth variation] In the above embodiment, the deterioration status of the piping member was predicted by using the frequency spectrum converted from the vibration acceleration detected by the sensor 100 as vibration evaluation data. In this modified example, the deterioration status of the piping member may be predicted by using the time waveform, which shows the change in vibration acceleration detected by the sensor 100 over time, as vibration evaluation data. The time waveform obtained in this way also exhibits characteristics unique to the deterioration status of the piping member. In this case, the machine learning model 2411 may be obtained by inputting learning data of the first to nth deterioration modes, which is formed by associating the time waveform obtained corresponding to the first to nth deterioration modes with the first to nth labels indicating that they are the first to nth deterioration modes, into a learning model creation device.

[0048] [Fifth variation] In the above embodiment, the deterioration status of the piping member was predicted by using the frequency spectrum converted from the vibration acceleration detected by the sensor 100 as vibration evaluation data. In this modified example, the deterioration status of the piping member may be predicted by using the natural frequency, which is the peak frequency of the frequency spectrum converted from the vibration acceleration detected by the sensor 100, as vibration evaluation data. The natural frequencies obtained in this way also exhibit characteristics unique to the deterioration status of the piping member. In this case, the machine learning model 2411 may be obtained by inputting learning data of the first to nth deterioration modes, which is formed by associating the first to nth labels indicating that the natural frequencies obtained corresponding to the first to nth deterioration modes with the first to nth deterioration modes, into a learning model creation device.

[0049] [Sixth variation] In each of the above embodiments, when predicting the quality of the piping installation, the worker applied vibration to the piping member in question by striking the position of the piping member corresponding to the joint to be predicted with a hammer. However, the source of vibration detected by the sensor in the piping component being predicted may not be from striking the piping component with a hammer or other striking device. For example, vibration may be applied to the piping component by striking it with a predetermined tool smaller than a hammer or with one's fingers. Furthermore, the vibration detected for predicting piping deterioration may not be vibration intentionally applied by an operator, but may be vibration caused by fluid flowing through the piping component, or ambient vibration transmitted from a piping component connected to the target of prediction at a certain distance, or from outside the piping. In this case, the vibration evaluation data in the aforementioned training data may also be created based on vibration acceleration obtained by detecting vibrations from vibration sources other than the hammer, similar to those described above.

[0050] [7th variation] In each of the embodiments described above, the prediction unit 232 was configured to predict the deterioration status of the piping components using a machine learning model. In this modified example, the prediction unit 232 is configured to predict the deterioration status of the piping components without using a machine learning model. In this case, the pipe deterioration prediction device 200 stores in the storage unit 204 a vibration evaluation data / deterioration status correspondence table that associates vibration evaluation data created for various deterioration statuses of pipe components with the corresponding deterioration status. The prediction unit 232 identifies the vibration evaluation data from among the vibration evaluation data stored in the vibration evaluation data / deterioration status correspondence table that most closely resembles the vibration evaluation data obtained based on the vibration acceleration detected by the sensor 100. The prediction unit 232 may output the deterioration status associated with the identified vibration evaluation data as the prediction result.

[0051] [8th variation] In Figure 1, an example is shown in which one sensor 100 is attached to a piping member. In this modified example, multiple sensors 100 may be attached to the piping member. For example, two pairs of sensors 100 may be mounted so that they are in a positional relationship where they face each other with respect to the center of the circle in the cross-section of the piping member. Furthermore, the conversion unit 231 may, for example, perform an operation to add or subtract the vibration acceleration signals of the two sensors 100 depending on whether the signal components corresponding to a specific vibration mode (for example, a crushing mode in which a pipe member vibrates as if it were being crushed) that is considered to have a high correlation with the deterioration status are in phase or out of phase in the vibration acceleration signals output from each of the two sensors 100. The conversion unit 231 may convert the calculated vibration acceleration signal into vibration evaluation data such as a frequency spectrum. In this case, since the above operation provides a vibration acceleration signal that emphasizes the specific vibration mode that is considered to have a high correlation with the deterioration status, the peak of the resonant frequency corresponding to the specific vibration mode that is considered to have a high correlation with the deterioration status is also emphasized in the converted frequency spectrum, and good prediction results can be expected. Alternatively, the conversion unit 231 may subtract or add the vibration acceleration signals of the two sensors 100 depending on whether the signal components corresponding to specific vibration modes (for example, a bending mode in which a pipe member flexes) that are considered to have a low correlation with the deterioration status are in opposite phase or in phase. The conversion unit 231 may convert the calculated vibration acceleration signals into vibration evaluation data such as a frequency spectrum. In this case, in the vibration acceleration signal obtained by the above calculation, the signal components of specific vibration modes that are considered to have a low correlation with the deterioration status are suppressed, and the signal components of specific vibration modes that are considered to have a relatively high correlation with the deterioration status are emphasized. As a result, in the frequency spectrum converted from the calculated vibration acceleration signal, the peak of the resonance frequency corresponding to the specific vibration mode that is considered to have a high correlation with the deterioration status is relatively emphasized, and good prediction results can be expected.

[0052] Furthermore, the processing of the pipe deterioration prediction device 200 may be performed by recording a program for realizing the functions of the pipe deterioration prediction device 200, etc., on a computer-readable recording medium, loading the program recorded on this recording medium into a computer system, and executing it. Here, "loading the program recorded on the recording medium into a computer system and executing it" includes installing the program into the computer system. Here, "computer system" includes hardware such as the OS and peripheral devices. Also, "computer system" may include multiple computer devices connected via a network including communication lines such as the Internet, WAN, LAN, and dedicated lines. Also, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into the computer system. Thus, the recording medium on which the program is stored may be a non-transient recording medium such as a CD-ROM. Also, the recording medium includes internal or external recording media that can be accessed from the distribution server for distributing the program. The program code stored on the distribution server's recording medium may be different from the program code in a format executable by the terminal device. In other words, the format in which the program is stored on the distribution server is irrelevant, as long as it can be downloaded from the distribution server and installed in an executable format on the terminal device. Furthermore, the program may be divided into multiple parts, each downloaded at a different time and then combined on the terminal device, and different distribution servers may distribute each of the divided programs. In addition, "computer-readable recording medium" includes volatile memory (RAM) within computer systems that act as servers or clients when a program is transmitted over a network, which retains the program for a certain period of time. Moreover, the program may only be used to implement some of the functions described above. Furthermore, the program may be a so-called differential file (differential program) that can implement the functions described above in combination with a program already recorded in the computer system.

[0053] <Note> (1) One aspect of this embodiment is a pipe deterioration prediction system comprising: a conversion unit (231) that converts vibration acceleration output from a sensor, which is provided to detect vibrations corresponding to vibrations occurring in a pipe member to be predicted in a pipe, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency; and a prediction unit (232) that predicts the deterioration status of the pipe member to be predicted based on the vibration evaluation data.

[0054] (2) One aspect of this embodiment is the pipe deterioration prediction system described in (1), wherein the prediction unit inputs the vibration evaluation data obtained by the conversion unit to a machine learning model which outputs the deterioration status of pipe members in response to the input of vibration evaluation data, and predicts the deterioration status of the pipe members to be predicted based on the deterioration status output by the machine learning model in response to the input of vibration evaluation data.

[0055] (3) One aspect of this embodiment is the pipe deterioration prediction system described in (2), wherein the machine learning model may be trained using training data that associates vibration evaluation data obtained based on vibrations occurring in each of a plurality of sample pipe members deteriorated by predetermined different deterioration conditions with deterioration conditions for the corresponding sample pipe members.

[0056] (4) One aspect of this embodiment is the pipe deterioration prediction system described in (3), wherein the deterioration condition includes the ultraviolet irradiation time.

[0057] (5) One aspect of this embodiment is a pipe deterioration prediction system as described in (3) or (4), wherein the machine learning model may output deterioration status corresponding to a predetermined condition value set as the deterioration condition, as well as deterioration status corresponding to a condition other than the condition value through interpolation or extrapolation.

[0058] (6) One aspect of this embodiment is a pipe deterioration prediction system according to any one of (3) to (5), comprising a machine learning model storage unit that stores a plurality of machine learning models corresponding to each different type of sample pipe member, each of which has been trained using the training data corresponding to each different type of sample pipe member, wherein the prediction unit determines the type of pipe member to be predicted based on the vibration evaluation data obtained by the conversion unit, and may predict the deterioration status using the machine learning model corresponding to the determined type from among the machine learning models stored in the machine learning model storage unit.

[0059] (7) One aspect of this embodiment is a pipe deterioration prediction system according to any one of (1) to (6), wherein the conversion unit may perform bandwidth limiting on vibration evaluation data as a frequency spectrum converted from the vibration acceleration.

[0060] (8) One aspect of this embodiment is the pipe deterioration prediction system described in (7), wherein the conversion unit may perform the bandwidth limiting on the vibration evaluation data as a frequency spectrum converted from the vibration acceleration such that frequency bands in which uncertainty occurs due to variability in the detection results by the sensor are removed.

[0061] (9) One aspect of this embodiment is the piping deterioration system described in (7) or (8), wherein the conversion unit may perform the bandwidth limiting on the vibration evaluation data as a frequency spectrum converted from the vibration acceleration such that a frequency band is removed in which a change occurs in accordance with the individual differences of the person striking the piping member to be predicted in order to generate vibration in the piping member to be predicted.

[0062] (10) One aspect of this embodiment is a pipe deterioration prediction system according to any one of (3) to (6), wherein the conversion unit performs bandwidth limiting on vibration evaluation data as a frequency spectrum converted from vibration acceleration, and the vibration evaluation data as a frequency spectrum included in the learning data may be subjected to the same bandwidth limiting as the vibration evaluation data as a frequency spectrum converted from vibration acceleration.

[0063] (11) One aspect of this embodiment is a pipe deterioration prediction method in a pipe deterioration prediction system, comprising: a conversion step in which a conversion unit converts vibration acceleration output from a sensor, which is provided to detect vibrations corresponding to vibrations occurring in a pipe member to be predicted in a pipe, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency; and a prediction step in which a prediction unit predicts the deterioration status of the pipe member to be predicted based on the vibration evaluation data.

[0064] (12) One aspect of this embodiment is a program that causes the computer in a pipe deterioration prediction system to function as a conversion unit that converts vibration acceleration output from a sensor, which is provided to detect vibrations corresponding to vibrations occurring in a pipe member to be predicted in the pipe, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency, and a prediction unit that predicts the deterioration status of the pipe member to be predicted based on the vibration evaluation data. [Explanation of Symbols]

[0065] 10 Actual piping, 20 Hammer, 100 Sensor, 200 Piping deterioration prediction device, 201 Data interface unit, 202 User interface unit, 203 Control unit, 204 Storage unit, 231 Conversion unit, 232 Prediction unit, 233 Output unit, 241 Machine learning model storage unit, 2411 Machine learning model

Claims

1. A conversion unit that converts vibration acceleration output from a sensor, which is installed to detect vibrations corresponding to vibrations occurring in the pipe components to be predicted in the piping, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency, A prediction unit predicts the deterioration status of the piping member to be predicted based on the vibration evaluation data, A pipe deterioration prediction system equipped with the following features.

2. The prediction unit inputs the vibration evaluation data obtained by the conversion unit into a machine learning model that outputs the deterioration status of piping members in response to the input of vibration evaluation data, and predicts the deterioration status of the target piping member based on the deterioration status output by the machine learning model in response to the input of vibration evaluation data. The pipe deterioration prediction system according to claim 1.

3. The machine learning model learns using training data that associates vibration evaluation data obtained based on vibrations generated in each of several sample piping members degraded under predetermined different degradation conditions with the degradation conditions applied to the corresponding sample piping members. The pipe deterioration prediction system according to claim 2.

4. The aforementioned degradation-inducing conditions include the irradiation time of ultraviolet light. The pipe deterioration prediction system according to claim 3.

5. The machine learning model outputs not only the degradation status corresponding to the predetermined condition value set as the degradation condition, but also degradation status corresponding to conditions other than the predetermined condition value through interpolation or extrapolation. The pipe deterioration prediction system according to claim 3 or 4.

6. The system includes a machine learning model storage unit that stores multiple machine learning models, each corresponding to a different type of sample piping component, which have been trained using the corresponding training data. The prediction unit determines the type of piping member to be predicted based on the vibration evaluation data obtained by the conversion unit, and predicts the deterioration status using the machine learning model corresponding to the determined type from among the machine learning models stored in the machine learning model storage unit. The pipe deterioration prediction system according to claim 3 or 4.

7. The conversion unit performs bandwidth limiting on the vibration evaluation data, which is converted from the vibration acceleration as a frequency spectrum. A pipe deterioration prediction system according to any one of claims 1 to 4.

8. The conversion unit performs the bandwidth limiting on the vibration evaluation data, which is a frequency spectrum converted from the vibration acceleration, so as to remove frequency bands where uncertainty arises due to variations in the detection results from the sensor. The pipe deterioration prediction system according to claim 7.

9. The conversion unit performs the bandwidth limiting on the vibration evaluation data, which is a frequency spectrum converted from the vibration acceleration, so as to remove frequency bands that would cause variations depending on the individual differences of the person striking the pipe member to be predicted in order to generate vibration in the pipe member to be predicted. The pipe deterioration prediction system according to claim 7.

10. The conversion unit performs bandwidth limiting on the vibration evaluation data, which is a frequency spectrum converted from the vibration acceleration. The vibration evaluation data as a frequency spectrum included in the aforementioned training data is subjected to the same bandwidth limitation as the vibration evaluation data as a frequency spectrum converted from the vibration acceleration. The pipe deterioration prediction system according to claim 3.

11. A method for predicting pipe deterioration in a pipe deterioration prediction system, The conversion unit performs a conversion step of converting vibration acceleration output from a sensor, which is provided to detect vibration acceleration corresponding to vibrations occurring in the pipe components to be predicted in the piping, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency. The prediction unit performs a prediction step of predicting the deterioration status of the piping member to be predicted based on the vibration evaluation data, A method for predicting pipe deterioration, including the following.

12. The computer in the pipe deterioration prediction system A conversion unit that converts vibration acceleration output from a sensor, which is installed to detect vibrations corresponding to vibrations occurring in the pipe components to be predicted in the piping, into vibration evaluation data in the form of a time waveform, frequency spectrum, or natural frequency. A prediction unit predicts the deterioration status of the target piping member based on the vibration evaluation data. A program designed to function as such.

Images