Facility inspection system and facility inspection method

Abstract

To provide a facility inspection system and a facility inspection method capable of easily determining the possibility of deterioration of a facility regardless of the size of the facility or the presence or absence of exterior materials.SOLUTION: A facility inspection system 100 comprises: an image acquisition device 10 which irradiates a facility 1 with an X-ray L0, detects a backscattered X-ray L1 being the X-ray L0 reflected by the facility 1, and acquires an X-ray image G based on the backscattered X-ray L1; a determination device 20 which has a machine learning model 22 learned with teacher data in which deterioration conditions of the facility 1 are linked with the X-ray image G, and determines presence or absence of a possibility of deterioration based on the X-ray image G by using the machine learning model 22; and an output device 30 which outputs a determination result calculated by the determination device 20.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an equipment inspection system and an equipment inspection method.

Background Art

[0002] Equipment such as pipes in a factory needs to be regularly inspected for deterioration such as whether corrosion has occurred. The pipe corrosion inspection device described in Patent Document 1 uses transmission X-rays to image the corrosion state of the pipe. Further, the corrosion state determination device described in Patent Document 2 uses a learned learning model to determine the corrosion state based on an image of a member of the equipment.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the pipe corrosion inspection method described in Patent Document 1, an X-ray generator for irradiating transmission X-rays and a sensor for detecting transmission X-rays must be installed on both sides of the equipment to be inspected, requiring a large working space. Further, when the plate thickness of the equipment is thick, the transmission X-rays cannot penetrate the equipment. Therefore, there is a risk that the pipe corrosion inspection method described in Patent Document 1 cannot inspect large equipment.

[0005] Further, in the corrosion state determination method described in Patent Document 2, when the equipment is covered with an exterior material such as a heat insulating material, the corrosion state of the equipment cannot be determined unless the exterior material is disassembled and removed. Since the installation of a temporary scaffold may be required to disassemble the exterior material, the corrosion state determination method described in Patent Document 2 has a problem that the time and cost required for equipment inspection become excessive.

[0006] The present invention has been made in view of such a situation, and an object thereof is to provide an equipment inspection system and an equipment inspection method capable of easily determining the possibility of deterioration of equipment regardless of the size of the equipment or the presence or absence of exterior materials.

Means for Solving the Problems

[0007] The inventor of the present invention conducted studies to achieve the above object, and found that by acquiring an X-ray image based on the backscattered X-ray reflected by the equipment to be inspected and determining the presence or absence of the possibility of deterioration based on the X-ray image using a machine learning model, the above object can be achieved, and the present invention has been completed.

[0008] That is, according to the present invention, the following equipment inspection system and equipment inspection method are provided.

[0009] [1] An equipment inspection system for determining the possibility of deterioration of equipment, comprising: an image acquisition device that irradiates the equipment with X-rays, detects backscattered X-rays that are the X-rays reflected by the equipment, and acquires an X-ray image based on the backscattered X-rays; a determination device having a machine learning model learned by teacher data in which the deterioration status of the equipment and the X-ray image are associated, and determining the presence or absence of the possibility of deterioration based on the X-ray image using the machine learning model; and an output device that outputs a determination result calculated by the determination device. [2] The equipment inspection system according to [1], wherein when the equipment is covered with an exterior material, the determination device determines the presence or absence of the possibility of deterioration of the equipment inside the exterior material. [3] The equipment inspection system according to [2], wherein the equipment is a pipe covered with the exterior material. [4] The equipment inspection system according to any one of [1] to [3], wherein the machine learning model determines the presence or absence of the possibility of rust, dew condensation, or scale generation in the equipment as the possibility of deterioration of the equipment. [5] The equipment inspection system according to any one of [1] to [4], wherein the machine learning model is trained using learning X-ray images obtained by imaging the equipment from a plurality of directions using the backscattered X-rays. [6] The equipment inspection system according to any one of [1] to [5], wherein the determination device is provided in a server. [7] An equipment inspection method for determining the possibility of equipment deterioration, comprising: irradiating the equipment with X-rays using an image acquisition device, detecting backscattered X-rays, which are the X-rays reflected by the equipment, and acquiring an X-ray image based on the backscattered X-rays; having a machine learning model trained with learning data in which the deterioration status of the equipment is associated with the feature amounts of the X-ray image using a determination device, and determining the presence or absence of the possibility of deterioration based on the X-ray image using the machine learning model; outputting the determination result calculated by the determination device to an output device.

Advantages of the Invention

[0010] According to the present invention, the possibility of equipment deterioration can be easily determined regardless of the size of the equipment or the presence or absence of an exterior material.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Embodiments for Carrying Out the Invention

[0012] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The facility inspection system 100 shown in FIG. 1 is a system for determining the presence or absence of the possibility of deterioration of facility 1. In the embodiment shown in FIG. 1, facility 1 is a metal pipe through which liquid or steam flows inside. Facility 1 is covered with an exterior material 2 on the outside. The exterior material 2 is formed of, for example, an iron sheet metal and functions as a heat insulating material. Note that facility 1 is not limited to a pipe and may be, for example, a tank of a distillation column.

[0013] When the facility inspection system 100 determines that rust S due to corrosion has occurred between facility 1 and exterior material 2, it determines that there is a possibility of deterioration in facility 1. Further, the determination device 20 is not limited to this, and when condensation or scale such as calcium carbonate has occurred between facility 1 and exterior material 2, it may determine that there is a possibility of deterioration.

[0014] The facility inspection system 100 includes an image acquisition device 10, a determination device 20, and an output device 30. Specifically, the image acquisition device 10 is a camera. Also, the determination device 20 is a controller of a server. The determination device 20 includes a feature extraction unit 21 and a machine learning model 22. Furthermore, the output device 30 is an electronic terminal. Information can be exchanged between the image acquisition device 10 and the determination device 20, and between the determination device 20 and the output device 30 through communication. Note that the present invention is not limited to this embodiment, and the image acquisition device 10, the determination device 20, and the output device 30 may be an integrated single device. Also, the output device 30 is not limited to an electronic terminal and may be, for example, a printer.

[0015] The image acquisition device 10 irradiates facility 1 with X-ray L0, detects the backscattered X-ray L1 reflected by facility 1, and acquires an X-ray image G based on the backscattered X-ray L1. The image acquisition device 10 transmits the acquired X-ray image G to the determination device 20.

[0016] Also, the feature extraction unit 21 of the determination device 20 extracts the features of the X-ray image G. The method of extracting the features of the X-ray image G will be specifically described with reference to FIG. 2. As shown in range A of FIG. 2, the greater the intensity of the backscattered X-ray L1 acquired by the image acquisition device 10, the higher the brightness appears in the X-ray image G. That is, the greater the intensity of the backscattered X-ray L1, the higher the whiteness in the monochrome X-ray image G. Note that the smaller the atomic number of the object irradiated with the X-ray L0, the greater the intensity of the backscattered X-ray L1. Here, when rust S has occurred between the facility 1 and the exterior material 2, in the range A where the rust S has occurred, the intensity of the backscattered X-ray L1 becomes greater than a predetermined value, so it is displayed as a whitish area in the X-ray image G. Similarly, when dew condensation or scale has occurred between the facility 1 and the exterior material 2, the intensity of the backscattered X-ray L1 with respect to the moisture or scale becomes greater than the predetermined value, so it is displayed as a whitish area in the X-ray image G. The feature extraction unit 21 extracts, as feature amounts, the area ratio (the ratio of the number of pixels occupied by the area A (the area with a brightness of a predetermined value or more) in the X-ray image G to the total number of pixels), the average value of the brightness of the area, the distribution, and the like.

[0017] Next, the machine learning model 22 of the determination device 20 shown in FIG. 1 determines the presence or absence of a deterioration possibility based on the X-ray image G. Specifically, the machine learning model 22 uses the feature amounts extracted by the feature extraction unit 21 as input data and outputs the presence or absence of a deterioration possibility by class classification.

[0018] Note that the present invention is not limited to this embodiment, and the determination device 20 may not have the feature extraction unit 21. That is, the machine learning model 22 may be a convolutional neural network, and may output the presence or absence of a deterioration possibility using the X-ray image G as input data.

[0019] Further, the determination device 20 may extract an area on the X-ray image G where rust S, dew condensation, or scale may have occurred, and calculate which of rust S, dew condensation, or scale the occurrence event in each area is classified into. Further, the determination device 20 may output the possibility that rust S, dew condensation, or scale has occurred in the facility 1 as a probability such as "~%".

[0020] The determination device 20 transmits the calculated determination result to the output device 30. As a result, the user can confirm whether there is a possibility of deterioration in a predetermined inspection target area of the facility 1 via the display of the output device 30. Further, the output device 30 may display, in color on the X-ray image G, locations where rust S, condensation, or scale may have occurred.

[0021] Next, with reference to FIG. 3, the learning method of the machine learning model 22 will be described in detail. The machine learning model 22 is learned using teacher data in which the deterioration status of the facility 1 and the feature amounts of the X-ray image G are associated. Specifically, as shown in FIG. 3, the machine learning model 22 is learned using teacher data generated by the teacher data generation device 40. The teacher data generation device 40 generates teacher data based on a plurality of data sets D that are combinations of the X-ray image G and the inspection results of the facility 1.

[0022] Here, the method for creating the data set D will be described. In the example shown in FIG. 3, 99 data sets D including inspection results 01 to 99 are created, but the number of data sets D is not limited to this. First, a predetermined area of the facility 1 is imaged from a plurality of directions (for example, four directions) using the image acquisition device 10. That is, for one area, a plurality of learning X-ray images Gt imaged from different directions are acquired. Then, for a plurality of areas, imaging is similarly repeated a plurality of times, and a plurality of learning X-ray images Gt (99 images in the example shown in FIG. 3) are collected. Next, in each of the imaged areas, the exterior material 2 of the facility 1 is removed, and data (inspection results 01 to 99) regarding the deterioration status of the surface of the facility 1 is acquired. The inspection results include the presence or absence of moisture, the water content range, the presence or absence of rust S, the dimensions of rust nodules, the presence or absence of scale, and the dimensions of the scale on the surface of the facility 1.

[0023] When the dataset D is input to the teacher data generation device 40, the teacher data generation device 40 extracts the feature amounts of each of the plurality of learning X-ray images Gt in the dataset D. Further, the teacher data generation device 40 performs labeling indicating the presence or absence of the possibility of deterioration for each piece of feature amount data based on the inspection results 01 to 99. For example, when moisture is present on the surface of the facility 1 and the water-containing range occupies a predetermined range or more, the teacher data generation device 40 assigns a label of "possibility of deterioration" to the corresponding feature amount data. Also, when rust S or scale is present on the surface of the facility 1 and the dimensions of the rust S or scale are equal to or greater than a predetermined value, the teacher data generation device 40 assigns a label of "possibility of deterioration" to the corresponding feature amount data. The labeled feature amount data is used as teacher data for learning the machine learning model 22. Note that, based on the inspection results 01 to 99, the user may manually perform labeling for each of the plurality of learning X-ray images Gt. That is, the teacher data used for learning the machine learning model 22 is data in which the deterioration state of the facility 1 and the X-ray image G are associated.

[0024] As described above, the facility inspection system 100 according to the present embodiment includes an image acquisition device 10 that irradiates the facility 1 with X-rays L0, detects backscattered X-rays L1 that are the X-rays reflected by the facility 1, and acquires an X-ray image G based on the backscattered X-rays L1. Further, the facility inspection system 100 has a machine learning model 22 learned from learning data in which the deterioration state of the facility 1 and the X-ray image G are associated, and includes a determination device 20 that determines the presence or absence of a deterioration possibility based on the X-ray image G using the machine learning model 22. Furthermore, the facility inspection system 100 includes an output device 30 that outputs the determination result calculated by the determination device 20. Thus, since the facility inspection system 100 determines the deterioration possibility of the facility 1 based on the X-ray image G acquired using the backscattered X-rays L1, even when the surface of the facility 1 is covered with the exterior material 2 and is difficult to visually recognize, the presence or absence of the deterioration possibility of the facility 1 can be determined. Also, if transmission X-rays are used for the inspection of the facility 1, a radiation device for emitting the transmission X-rays and a sensor for detecting the transmission X-rays or a film for absorbing the transmission X-rays need to be installed on both sides of the facility 1, respectively. In contrast, since the facility inspection system 100 according to the present embodiment uses the backscattered X-rays L1, the facility 1 only needs to be imaged from one direction. As a result, the work space for inspecting the facility 1 can be reduced, and the inspection work of the facility 1 can be simplified. Also, since the facility inspection system 100 uses the backscattered X-rays L1, the X-ray image G can be acquired regardless of the plate thickness of the facility 1. Therefore, the facility inspection system 100 can easily determine the deterioration possibility of the facility 1 even if the facility 1 is a large facility such as a tank of a distillation column. Also, since the facility inspection system 100 determines the presence or absence of a deterioration possibility using the machine learning model 22, the deterioration possibility of the facility 1 can be determined with high accuracy without being affected by the noise included in the X-ray image G. Also, by using the facility inspection system 100, the user can inspect the deterioration state of the facility 1 with a predetermined or higher accuracy regardless of experience and knowledge.

[0025] In addition, when the facility 1 is covered with the exterior material 2, the determination device 20 of the facility inspection system 100 determines the presence or absence of the possibility of deterioration of the facility 1 inside the exterior material 2. Thereby, even when the facility 1 is covered with the exterior material 2, the facility inspection system 100 can determine the possibility of deterioration of the surface of the facility 1 without disassembling the exterior material 2 by using the backscattered X-ray L1.

[0026] In addition, the facility 1 inspected by the facility inspection system 100 is a pipe covered with the exterior material 2. Thereby, the facility inspection system 100 can determine the possibility of deterioration of the pipe without removing the exterior material 2. Further, since the facility inspection system 100 has the machine learning model 22, even when the X-ray image G contains noise due to the influence of the fluid flowing inside the pipe, the possibility of deterioration of the facility 1 can be determined with high accuracy without being affected by the noise.

[0027] In addition, the machine learning model 22 of the facility inspection system 100 determines the presence or absence of the possibility of rust S, condensation, or scale generation in the facility 1 as the possibility of deterioration of the facility 1. Thereby, the facility inspection system 100 can easily determine whether rust S, condensation, or scale has occurred in the facility 1.

[0028] In addition, the machine learning model 22 is trained using the learning X-ray image Gt obtained by imaging the facility 1 from a plurality of directions using the backscattered X-ray L1. Thereby, the facility inspection system 100 can improve the determination accuracy of the possibility of deterioration of the facility 1.

[0029] In addition, the determination device 20 is provided in the server. Thereby, the user can confirm the determination result of the possibility of deterioration of the facility 1 calculated by the determination device 20 on the server side via the output device 30 located at a place away from the installation location of the determination device 20.

Explanation of Signs

[0030] 100... Facility inspection system 1... Facility 2... Exterior material 10... Image acquisition device 20... Determination device 22…Machine learning model 30…Output device G…X-ray image Gt…X-ray image for training L1…Backscattered X-ray

Claims

1. An equipment inspection system for determining the possibility of deterioration of equipment, comprising: an image acquisition device that irradiates the equipment with X-rays, detects backscattered X-rays that are the X-rays reflected by the equipment, and acquires an X-ray image based on the backscattered X-rays; a determination device having a machine learning model learned from teacher data in which the deterioration status of the equipment and the X-ray image are associated, and determining the presence or absence of the possibility of deterioration based on the X-ray image using the machine learning model; and an output device that outputs a determination result calculated by the determination device.

2. The equipment inspection system according to claim 1, wherein when the equipment is covered with an exterior material, the determination device determines the presence or absence of the possibility of deterioration of the equipment inside the exterior material.

3. The equipment inspection system according to claim 2, wherein the equipment is a pipe covered with the exterior material.

4. The equipment inspection system according to claim 1, wherein the machine learning model determines the presence or absence of the possibility of rust, condensation, or scale generation in the equipment as the possibility of deterioration of the equipment.

5. The equipment inspection system according to claim 1, wherein the machine learning model is learned using learning X-ray images obtained by imaging the equipment from a plurality of directions using the backscattered X-rays.

6. The equipment inspection system according to claim 1, wherein the determination device is provided in a server.

7. An equipment inspection method for determining the possibility of deterioration of equipment, comprising: using an image acquisition device to irradiate the equipment with X-rays, detect backscattered X-rays that are the X-rays reflected by the equipment, and acquire an X-ray image based on the backscattered X-rays; using a determination device to have a machine learning model learned from learning data in which the deterioration status of the equipment and the feature amount of the X-ray image are associated, and determining the presence or absence of the possibility of deterioration based on the X-ray image using the machine learning model; and outputting the determination result calculated by the determination device to an output device.

Images