KR102601916B1 - Image base pipe damage detecting system and method for detecting pipe damage using the same - Google Patents

Abstract

In the image-based piping damage detection system and the piping damage detection method using the same, the piping damage detection system includes a piping system, an imaging unit, a learning unit, and a detection unit. The piping system includes a plurality of pipes, a plurality of valves provided in the pipes, and a pressure sensor provided in each of the valves. The imaging unit converts the pressure data of each valve measured by the pressure sensor into an image. The learning unit learns damage to the piping system based on the converted image. The detection unit detects damage to the piping system based on the learning results of the learning unit from the input pressure data of each valve.

Description

Image-based pipe damage detection system and pipe damage detection method using the same {IMAGE BASE PIPE DAMAGE DETECTING SYSTEM AND METHOD FOR DETECTING PIPE DAMAGE USING THE SAME}

The present invention relates to a piping damage detection system and a piping damage detection method using the same. More specifically, when piping at a specific location in a piping system designed in various ways is damaged, pressure data around valves installed in the piping system are converted into images. , It is about an image-based pipe damage detection system that can accurately detect the location of damaged pipes based on learning, and a pipe damage detection method using the same.

In complex piping systems, various technologies are being developed to accurately detect damage to pipes at specific locations.

Republic of Korea Patent No. 10-2290217 discloses a technology that receives sensing signals from a plurality of valves provided in a pipe, determines the location of damage, and controls the operation. In this case, the transmitted sensing signal is digital communication, Communication means such as analog communication or PLC communication are used.

In addition, Republic of Korea Patent No. 10-1915638 discloses technology related to a multi-channel valve control and self-failure diagnosis system that remotely controls and monitors the operation status of the valve, providing operation status and maintenance information, as well as failure prevention.

Furthermore, Republic of Korea Patent No. 10-2392595 discloses a technology related to a pipe damage detection method that determines pipe damage based on pressure sensing information from a pressure sensor.

However, in the case of the above-described prior technologies, it is characterized in that damage is determined using a sensing signal measured through a valve. In particular, the measurement results of the pressure sensor show that the pressure difference between valves installed in the piping system is similar, indicating damage. There is a problem with the low accuracy of judgment.

Republic of Korea Patent No. 10-2290217 Republic of Korea Patent No. 10-1915638 Republic of Korea Patent No. 10-2392595

Accordingly, the technical problem of the present invention was conceived from this point, and the purpose of the present invention is to convert pressure data around the valves installed in the piping system into images when the piping at a specific location in the variously designed piping system is damaged, thereby enabling learning. Based on this, we provide an image-based pipe damage detection system that can accurately detect the location of damaged pipes.

Additionally, another object of the present invention is to provide a method for detecting pipe damage using the pipe damage detection system.

A piping damage detection system according to an embodiment for realizing the object of the present invention described above includes a piping system, an imaging unit, a learning unit, and a detection unit. The piping system includes a plurality of pipes, a plurality of valves provided in the pipes, and a pressure sensor provided in each of the valves. The imaging unit converts the pressure data of each valve measured by the pressure sensor into an image. The learning unit learns damage to the piping system based on the converted image. The detection unit detects damage to the piping system based on the learning results of the learning unit from the input pressure data of each valve.

In one embodiment, the pressure sensor may include an inlet pressure sensor that measures the pressure of the inlet of each valve, and an outlet pressure sensor that measures the pressure of the outlet of each valve.

In one embodiment, it may further include a calculation unit that calculates characteristic factors based on pressure data (P1) of the inlet pressure sensor and pressure data (P2) of the outlet pressure sensor.

In one embodiment, the calculated characteristic factor is the sum of P1 and P2 (P1+P2), the difference between P1 and P2 (P1-P2), the product of P1 and P2 (P1*P2), and the division of P1 and P2. (P1/P2), maximum of P1 and P2 (max(P1, P2)), minimum of P1 and P2 (min(P1, P2)), square of P1 (P1 ² ), square of P2 (P2 ² ) It may be defined as any one of these, or a combination of at least one or more of them.

In one embodiment, the detection unit detects damage to the piping system based on learning results learned using all of the defined characteristic factors, or, among the results learned through the defined characteristic factors, the learning result is the best. Damage to the piping system can be detected based on accurately derived characteristic factors.

In one embodiment, the imaging unit may convert the calculated characteristic factor into an image.

In one embodiment, the calculation unit may calculate the characteristic factors in time series and provide each calculation result to the imaging unit.

In one embodiment, the imaging unit may assign the calculated characteristic factor to any one pixel value between 0 and 255 and convert it into an image.

In one embodiment, the learning unit may learn about damage to the piping system by learning images of valves around damaged pipes and images of valves around undamaged pipes.

A pipe damage detection method according to an embodiment for realizing another object of the present invention described above includes a plurality of pipes, a plurality of valves provided in the pipes, and a pressure sensor provided in each of the valves. With respect to a piping system including a step of learning about the damage state of the piping system, measuring the pressure of each valve at the pressure sensor while the piping system is in operation, converting the measured pressure data into an image. and detecting damage to the piping system from the converted image, based on the results of the learning.

In one embodiment, converting the measured pressure data into an image may include calculating the measured pressure data with predefined characteristic factors, and converting the calculated result into an image. .

In one embodiment, in the step of performing the learning, measuring the pressure of each valve by the pressure sensor with respect to the damaged or normal state of the piping system, a characteristic factor predefined based on the measured pressure data It may include calculating in time series, converting the calculated characteristic factors into an image, and learning the image depending on whether the piping system is damaged or normal.

In one embodiment, the step of measuring the pressure of each valve may include measuring the pressure at the inlet of each valve and the pressure at the outlet of each valve.

In one embodiment, the step of externally displaying the detected damage to the piping system may be further included.

According to embodiments of the present invention, the pipe damage location is detected based on the pressure information of the pressure sensor provided in the valve, and the damage location is detected by learning imaged data, so that the damage location is detected with high accuracy even in situations where the pressure difference is not large. Detection is possible and can be performed even in a variety of situations where the sensor malfunctions or is inoperable.

In other words, based on the previously learned learning results, it is possible to immediately check whether the piping system is damaged by converting time-series input pressure data into an image, enabling real-time damage detection as well as accurate damage detection. In addition, this real-time accurate damage detection can not only quickly shut off surrounding valves, but also enable rapid recovery of damaged pipes and minimize the occurrence of additional problems.

In particular, when performing learning, pressure data is calculated with predefined characteristic factors and imaged. By defining characteristic factors in various ways, images according to various characteristic factors can be derived, and through this, optimal damage determination can be made. This learning can be done.

In other words, learning is performed on all defined characteristic factors, and the learning results can be used by imaging all characteristic factors for pressure data to identify the input damage state, so that damage can be determined by defining characteristic factors in various ways. accuracy can be increased.

In contrast, learning is performed on all defined characteristic factors, the characteristic factors judged to be most accurate as a result of the learning are extracted, and calculations are performed on the pressure data to identify the input damage status using only the characteristic factors. By doing so, the calculation speed and accuracy for damage determination can be improved by using only the characteristic factors with the highest accuracy.

Figure 1 is a block diagram showing a pipe damage detection system according to an embodiment of the present invention.
Figure 2 is a schematic diagram showing an example of the piping system of Figure 1.
FIG. 3 is a flowchart showing a method of detecting pipe damage using the pipe damage detection system of FIG. 1.
FIG. 4 is a flowchart showing steps for performing learning in FIG. 3.
Figure 5 is a flowchart showing steps for converting the image of Figure 3.
FIG. 6A is a schematic diagram showing a case where damage occurs in the piping system of FIG. 2, and FIG. 6B is a graph illustrating a pressure measurement section for learning when damage occurs in FIG. 6A.
FIG. 7 is an image illustrating an image in a damaged state and an image in an undamaged state that are the object of learning in the step of performing the learning in FIG. 4 .

Since the present invention can be subject to various changes and can have various forms, embodiments will be described in detail in the text. However, this is not intended to limit the present invention to a specific disclosed form, and should be understood to include all changes, equivalents, and substitutes included in the spirit and technical scope of the present invention. While describing each drawing, similar reference numerals are used for similar components. Terms such as first, second, etc. may be used to describe various components, but the components should not be limited by the terms.

The above terms are used only for the purpose of distinguishing one component from another. The terms used in this application are only used to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise.

In this application, terms such as “comprise” or “consist of” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the technical field to which the present invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and unless explicitly defined in the present application, should not be interpreted in an ideal or excessively formal sense. No.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the attached drawings.

Figure 1 is a block diagram showing a pipe damage detection system according to an embodiment of the present invention. Figure 2 is a schematic diagram showing an example of the piping system of Figure 1.

Referring to Figures 1 and 2, the pipe damage detection system 10 according to this embodiment includes a piping system 100, a calculation unit 200, an imaging unit 300, a learning unit 400, a detection unit 500, and a display unit. Includes (600).

First, as shown in FIG. 2, the piping system 100 is a piping system in which a plurality of valves 110 are provided on the piping 105, and is used to flow fluid into the piping 105. Drive pumps 101 and 102 may be provided at predetermined positions.

Furthermore, in this embodiment, a pressure sensor 120 including an inlet pressure sensor 121 and an outlet pressure sensor 122 must be provided to measure pressure data at the inlet and outlet of each of the valves 110. do.

At this time, the pressure sensor 120 is illustrated as being provided only at the inlet and outlet of a specific valve (SV10) through FIG. 2, but all valves (SV1 to SV10) provided in the piping system 100 of FIG. 2 ) Must be provided at each inlet and outlet.

Additionally, since the piping system 100 can be designed in various ways, it is not limited to a specific structure. Accordingly, Figure 2 also shows an example of the configuration of the piping system 100, and corresponds to the piping system illustrated for convenience of explanation.

That is, as illustrated in FIG. 2, in the piping system 100, first to tenth valves SV1 to SV10 may be provided on the piping 105 configured as a parallel circuit, and the first to tenth valves SV1 to SV10 may be provided on the piping 105 configured as a parallel circuit. The locations of the tenth valves may vary, but each may be spaced apart from each other by a predetermined distance and may be generally uniformly provided on the pipe 105.

Furthermore, in the case of the calculation unit 200, imaging unit 300, learning unit 400, detection unit 500, and display unit 600, which constitute the pipe damage detection system 10, for convenience of explanation, the pipe damage will be described later. This is explained in detail through an explanation of the damage detection method.

FIG. 3 is a flowchart showing a method of detecting pipe damage using the pipe damage detection system of FIG. 1. Figure 4 is a flowchart showing steps for performing the learning of Figure 3. Figure 5 is a flowchart showing steps for converting the image of Figure 3. FIG. 6A is a schematic diagram showing a case where damage occurs in the piping system of FIG. 2, and FIG. 6B is a graph illustrating a pressure measurement section for learning when damage occurs in FIG. 6A. FIG. 7 is an image illustrating an image in a damaged state and an image in an undamaged state that are the object of learning in the step of performing the learning in FIG. 4 .

First, referring to FIG. 3, in the piping damage detection method, the learning unit 400 learns about the damage state of the piping system 100 (step S10).

At this time, learning in the learning unit 400 is performed for various situations such as when the piping system 100 is in a normal state without damage and in a damaged state.

Meanwhile, damage to the piping system 100 means that the piping 105 constituting the piping system 100 is damaged. If the piping 105 is damaged at a specific location, the damaged piping 105 The pressure of the surrounding valves drops rapidly.

An example of this damage condition is illustrated in Figure 6A. That is, in a damaged state of the piping system 100 as shown in FIG. 6A, a rupture occurs in the piping between the first valve SV1 and the fourth valve SV4, and the seventh valve SV7 and the fourth valve SV4 are damaged. This is a case where damage (rupture) occurs in the pipe between valve 10 (SV10).

Therefore, based on the pressure change of surrounding valves due to pipe damage, a certain amount of learning, which will be described later, is performed.

More specifically, referring to FIG. 4 at the same time, when performing learning in the learning unit 400 (step S10), first, the pressure of each valve (SV1 to SV10) is measured by the pressure sensor 120. (Step S11).

That is, the pressure sensor 120 measures the pressure of each valve (SV1 to SV10) assuming various damaged states as well as a normal state. In this case, various damage states mean states in which pipes constituting the piping system are damaged at various locations. In addition, the location of damage to the pipe may be at any one location, and damage at multiple locations is also possible.

Scenarios for these normal states and various damaged states can be preset and stored, and the learning unit 400 performs learning on the scenarios for various preset states.

Meanwhile, as described above, the pressure sensor 120 is individually provided in each of the valves SV1 to SV10, and is provided at the inlet and outlet of each valve. Therefore, the measured pressure data includes inlet pressure data (P1) measured at the inlet pressure sensor 121 of each valve (SV1 to SV10), and outlet pressure sensor ( Includes outlet pressure data (P2) measured at 122).

As described above, when inlet pressure data (P1) and outlet pressure data (P2) are measured for each valve, the pressure drops in a specific valve according to each damage state, so each valve is measured according to each damage scenario. The pressure data for the valves in is variable.

Afterwards, when the pressure data is acquired (step S11), the calculation unit 200 calculates predefined characteristic factors in time series based on the measured pressure data (step S12).

That is, in this embodiment, the learning is not performed using the measured pressure data as is, but the calculation is primarily performed using predefined characteristic factors.

At this time, the characteristic factor may be predefined as a factor derived by performing a predetermined operation on the inlet pressure data (P1) and outlet pressure data (P2) measured for each valve.

The characteristic factors predefined in this way are, for example, the sum of P1 and P2 (P1+P2), the difference between P1 and P2 (P1-P2), the product of P1 and P2 (P1*P2), and the sum of P1 and P2 (P1+P2). Divide by (P1/P2), the largest of P1 and P2 (max(P1, P2)), the smallest of P1 and P2 (min(P1, P2)), the square of P1 (P1 ² ), the square of P2 (P2 ² ) may be any one of the following.

Furthermore, in addition to each of the above-defined feature factors, that is, one feature factor, a combination of at least one of the above-defined feature factors may be defined as one feature factor. In other words, the characteristic factor may be defined as two factors, (P1+P2) and (P1-P2), and the characteristic factor may be defined as three factors, (P1+P2), (P1-P2), and (P1*P2). It can also be defined as .

Ultimately, the calculation unit 200 derives values defined as characteristic factors for the inlet pressure data (P1) and outlet pressure data (P2) based on the characteristic factors defined as above.

Furthermore, the calculation unit 200 performs the derivation of the characteristic factors described above in a time-series manner. In this case, performing the calculation in time series means deriving the value defined by the characteristic factor for each hour as time passes at specific time intervals. To achieve this, it is obvious that the inlet pressure data (P1) and outlet pressure data (P2) must be measured in advance from the pressure sensor 120 every hour.

That is, as shown in FIG. 6B, when damage to a valve occurs in the piping system 100 illustrated in FIG. 6A, the pressure of all valves (SV1 to SV10) of the piping system 100 immediately drops immediately after damage (A ). At this time, as the valves that sense this pressure drop constantly change the degree of opening, the pressure damage of each valve (SV1 to SV10) changes depending on the position of the valve over time (B). After this, when the valves (SV, SV4, SV7, SV10) close to the damaged pipe determine that there is damage in the vicinity and block it, the pressure of the other valves returns to the initial level. You will recover (C).

As described above, when damage occurs in a pipe at a specific location in the piping system 100, the pressure of each valve SV1 to SV10 changes depending on the location of damage over time.

Accordingly, the calculation unit 200 performs the derivation of the above-described characteristic factors in a time-series manner, so that more accurate learning of the pressure change of each valve according to the damage state can be performed.

In addition, in this case, when performed in time series, the measured pressure section starts from the point when the pressure drops in all valves as soon as damage occurs, as shown in Figure 6b, and after the valves start controlling the degree of opening, a certain period of time elapses. It can be defined as a point in time afterward. In other words, it can be defined as the point at which each valve closes to a preset specific angle through control of the degree of opening.

Accordingly, the derivation of the time series characteristic factors can be performed only for the defined pressure section.

Afterwards, the imaging unit 300 converts the characteristic factors calculated by the calculation unit 200 into a predetermined image (step S13).

At this time, conversion to an image can be performed in various ways, but for example, the calculation result of the calculated characteristic factor can be converted to an image by assigning it to any one pixel value between 0 and 255. .

For example, for a characteristic factor defined as P1+P2, the result value of calculating P1+P2 can be converted into an image by assigning a pixel value between 0 and 255, and at this time, the time-series data is displayed according to the pipe damage state. Since P1+P2 have different operation result values, they can be assigned different pixel values and converted into different images.

It is clear that, when a plurality of characteristic factors are defined, the image conversion as described above must be performed in time series for each of the plurality of characteristic factors.

Figure 7 illustrates an image resulting from conversion of these characteristic factors into an image by the imaging unit 300.

That is, as shown in FIG. 7, for each of the first to tenth valves (SV1 to SV10) (horizontal axis arrangement in FIG. 7), images converted to characteristic factors are generated in time series (vertical axis arrangement in FIG. 8). It can be derived.

Afterwards, the learning unit 400 performs learning on the converted image (step S14).

The converted image has already been defined to which scenario the image corresponds. Therefore, the learning unit 400 determines whether the piping system 100 is in a normal state as well as in a damaged state, and if it is damaged, at which location the piping is damaged. You can learn about characteristics.

That is, as shown in FIG. 7, the learning unit 400 performs learning on images 510 of valves 501 around which pipe damage exists as images in a damaged state, and For valves 502 that do not have damage in the vicinity, learning is performed using images in a normal state.

In addition, since this learning is performed on time-series images, the learning unit 400 can also learn how images in a damaged state and images in a normal state change in time series.

This performance in the learning unit 400 may be performed using a machine learning method for images.

As described above, the learning unit 400 performs learning on the damage state of the piping system 100 defined by various scenarios (step S10).

In addition, the learning of the learning unit 400 is performed in advance, and it is possible to detect whether the piping system is damaged through the learning unit 400 that has performed such learning, as will be described later.

That is, when the piping system 100, which requires detection of actual damage, is operated, the pressure of each valve (SV1 to SV10) is measured by the pressure sensor 120 described above while the piping system 100 is operating. (step S20).

The pressure measurement of each valve by the pressure sensor 120 is substantially the same as the pressure measurement step (step S11) of each valve by the pressure sensor in the learning step described above.

Accordingly, the inlet pressure sensor 121 and the outlet pressure sensor 122 of the pressure sensor 120 measure the pressures at the inlet and outlet of each valve SV1 to SV10, respectively.

Afterwards, the imaging unit 300 converts the inlet and outlet pressure data of each measured valve (SV1 to SV10) into an image (step S30).

More specifically, in the step of converting into such an image (step S30), as shown in FIG. 5, the pressure data measured by the pressure sensor 120 is calculated with predefined characteristic factors (step S31).

In this case, the characteristic factors are the sum of P1 and P2 (P1+P2), the difference between P1 and P2 (P1-P2), the product of P1 and P2 (P1*P2), and the sum of P1 and P2, as exemplified above. Divide by (P1/P2), the largest of P1 and P2 (max(P1, P2)), the smallest of P1 and P2 (min(P1, P2)), the square of P1 (P1 ² ), the square of P2 (P2 ² ), or a combination of at least one or more.

Additionally, the calculation result of the characteristic factor is converted into an image through the imaging unit 300 (step S32). Since the image conversion in the imaging unit 300 is the same as described in the previous learning step (step S13), redundant description will be omitted.

As described above, when the measured pressure data is converted into an image, the detection unit 500 uses the converted image to determine the operating piping system 100 based on the results of learning in the learning unit 400. Damage is detected (step S40).

That is, when the converted image is provided to the learning unit 400, the learning unit 400 compares the provided image with the image from the existing learning results for each damage scenario, and corresponds to the damage state. It is determined whether damage occurs, and if it is in a damaged state, at what location in the piping system 100 the damage occurred. The determined result is provided to the detection unit 500, and damage to the piping system is detected. .

Meanwhile, in the learning process, the learning unit 400 learns the characteristics of images in normal and damaged states for all predefined characteristic factors.

However, when determining whether there is actual damage, the learning results for all characteristic factors in the learning unit 400 may not be used.

That is, during the learning process in the learning unit 400, characteristic factors that can lead to the most accurate results in determining the normal state and the damaged state can be extracted, and accordingly, in determining whether there is actual damage. , damage to the piping system 100 can be detected by using only the characteristic factors with the highest accuracy in the determination.

Although learning is performed on all of the various characteristic factors illustrated above, for example, the most accurate learning method is to judge the normal state and the damaged state based on the maximum value (max(P1, P2)) of P1 and P2. If the results are derived, in detecting the damage state of the piping system in actual operation, detection can be performed using only the maximum value (max(P1, P2)) of P1 and P2 as a characteristic factor.

In this case, it is sufficient to calculate only max(P1, P2) in the characteristic parameter calculation of the calculation unit 200, and the result of the learning in the learning unit 400 as well as the image conversion in the imaging unit 300 Even when using , only the characteristic factors max(P1, P2) can be utilized, thereby minimizing the overall data processing speed and enabling rapid detection of pipe damage.

Of course, it is also possible to detect pipe damage using all defined characteristic factors, rather than selecting only specific characteristic factors that produce the most accurate results. In this way, in the case of detecting pipe damage using all characteristic factors, the accuracy of specific characteristic factors is not high, especially in the learning results of the learning unit 400, and the judgment result is determined by applying a plurality of specific factors simultaneously. It can be said to be particularly effective when accuracy is high.

As described above, the detection unit 500 can use the learning results of the learning unit in various ways when detecting damage to a piping system currently in operation.

In addition, the piping system damage detection result of the detection unit 500 is displayed externally through a separate display unit 600 (step S50), and necessary tasks can be performed accordingly.

That is, although not shown, the piping system 100 may further include a control unit that controls the opening and closing of each valve SV1 to SV10 based on the damage state.

Accordingly, the control unit may control the closing operation of valves close to the location where the damage occurred, based on the detection result of the damage state of the piping system 100 by the detection unit 500.

For example, through the detection result of the detection unit 500, as shown in FIG. 6A, in the piping system 100, between the first valve (SV1) and the fourth valve (SV4), and the seventh valve (SV7) If a pipe rupture occurs between the tenth valve SV10, the control unit controls valves located around the pipe damage point to close.

That is, the first valve (SV1), the fourth valve (SV4), the seventh valve (SV7), and the tenth valve (SV10) are closed, and the other valves are open (open). ) status is maintained.

In addition, through the control unit's opening and closing control of the valves, as shown in FIG. 6b, the pressure returns to the initial pressure in all valves except those close to the damaged position (C), Through this, the piping system 100 is prevented from leaking due to damage, enabling stable operation in other areas except for the damaged area.

According to the embodiments of the present invention as described above, the pipe damage location is detected based on the pressure information of the pressure sensor provided in the valve, and the damage location is detected by learning the imaged data, so even in situations where the pressure difference is not large. Detection is possible with high accuracy and can be performed even in a variety of situations where the sensor is malfunctioning or inoperable.

In other words, based on the previously learned learning results, it is possible to immediately check whether the piping system is damaged by converting time-series input pressure data into an image, enabling real-time damage detection as well as accurate damage detection. In addition, this real-time accurate damage detection can not only quickly shut off surrounding valves, but also enable rapid recovery of damaged pipes and minimize the occurrence of additional problems.

In particular, when performing learning, pressure data is calculated with predefined characteristic factors and imaged. By defining characteristic factors in various ways, images according to various characteristic factors can be derived, and through this, optimal damage determination can be made. This learning can be done.

In other words, learning is performed on all defined characteristic factors, and the learning results can be used by imaging all characteristic factors for pressure data to identify the input damage state, so that damage can be determined by defining characteristic factors in various ways. accuracy can be increased.

In contrast, learning is performed on all defined characteristic factors, the characteristic factors judged to be most accurate as a result of the learning are extracted, and calculations are performed on the pressure data to identify the input damage status using only the characteristic factors. By doing so, the calculation speed and accuracy for damage determination can be improved by using only the characteristic factors with the highest accuracy.

Although the present invention has been described above with reference to preferred embodiments, those skilled in the art can make various modifications and changes to the present invention without departing from the spirit and scope of the present invention as set forth in the following patent claims. You will understand that it is possible.

10: Piping damage detection system 100: Piping system
110: valve 120: pressure sensor
121: Inlet pressure sensor 122: Outlet pressure sensor
200: calculation unit 300: imaging unit
400: learning unit 500: detection unit
600: display unit

Claims (14)

A piping system including a plurality of pipes, a plurality of valves provided in the pipes, and a pressure sensor provided in each of the valves;
A calculation unit that calculates characteristic factors based on the inlet pressure data (P1) and outlet pressure data (P2) of each valve;
an imaging unit that converts the calculated characteristic factors into an image;
a learning unit that learns damage to the piping system based on the converted image; and
A detection unit that detects damage to the piping system based on the learning results of the learning unit from the input pressure data of each valve,
The calculation unit derives the characteristic factors every hour, and calculates the characteristic factors in time series for a predetermined period of time from the time the damage occurs. The method of claim 1, wherein the pressure sensor is:
an inlet pressure sensor that measures the pressure at the inlet of each valve; and
A pipe damage detection system comprising an outlet pressure sensor that measures the pressure at the outlet of each valve. delete The method of claim 1, wherein the calculated characteristic factor is:
Sum of P1 and P2 (P1+P2), difference between P1 and P2 (P1-P2), product of P1 and P2 (P1*P2), division of P1 and P2 (P1/P2), maximum value between P1 and P2 (max(P1, P2)), the minimum value of P1 and P2 (min(P1, P2)), any one of the square of P1 (P1 ² ), the square of P2 (P2 ² ), or a combination of at least one of these. A pipe damage detection system characterized in that it is defined. The method of claim 4, wherein the detection unit,
Detect damage to the piping system based on learning results learned using all of the defined characteristic factors, or
A piping damage detection system characterized in that it detects damage to the piping system based on the characteristic factor that most accurately derives the learning result among the results learned through the defined characteristic factors. delete The method of claim 1, wherein the calculation unit,
A pipe damage detection system characterized by calculating the characteristic factors in time series and providing each calculation result to the imaging unit. The method of claim 1, wherein the imaging unit:
A pipe damage detection system characterized in that the calculated characteristic factors are assigned to any one pixel value between 0 and 255 and converted into an image. The method of claim 8, wherein the learning unit,
A pipe damage detection system characterized by learning damage to the piping system by learning images of valves around damaged pipes and images of valves around undamaged pipes. Regarding a piping system including a plurality of pipes, a plurality of valves provided in the pipes, and a pressure sensor provided in each of the valves,
Performing learning about the damage state of the piping system;
With the piping system in operation, measuring the pressure of each valve using the pressure sensor;
Calculating characteristic factors based on the inlet pressure data (P1) and outlet pressure data (P2) of each valve;
converting the calculated characteristic factor into an image; and
A step of detecting damage to the piping system from the converted image, based on the results of the learning,
The characteristic factor is a pipe damage detection method characterized by deriving the characteristic factor every hour and calculating the characteristic factor in time series for a predetermined period of time from the time the damage occurs. delete The method of claim 10, wherein in performing the learning,
Regarding the state in which the above piping system is damaged or normal,
Measuring the pressure of each valve using the pressure sensor;
Calculating predefined characteristic factors in time series based on the measured pressure data;
converting the calculated characteristic factor into an image; and
A pipe damage detection method comprising the step of learning the image according to whether the pipe system is damaged or normal. delete According to clause 10,
A piping damage detection method further comprising displaying the detected damage to the piping system externally.