Pump inspection report generation method and pump inspection report generation support system

The method and system address the issue of unsuitable inspection reports by using a fiberscope and AI to extract and process images for both visual inspection and image processing, resulting in user-friendly and efficient pump inspection reports.

JP2024168143APending Publication Date: 2024-12-05HITACHI IND PROD LTD
View PDF 17 Cites 0 Cited by

Patent Information

Application Number
JP2023084577
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-23
Publication Date
2024-12-05

AI Technical Summary

Technical Problem

Conventional pump inspection systems fail to generate inspection reports suitable for users, as they do not adequately consider the difference between images for visual inspection and those suitable for image processing, often resulting in unsuitable reports.

Method used

A method and system that extracts images suitable for both image processing and visual inspection, using a fiberscope to acquire images, and incorporates AI for automated report generation, including a captured image management unit, an aging comparison image search, and a report output unit to create user-friendly reports.

Benefits of technology

Enables the automatic or semi-automatic creation of inspection reports tailored to user needs, improving the accuracy and efficiency of pump maintenance by integrating image processing and visual inspection techniques.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024168143000001_ABST
    Figure 2024168143000001_ABST
Patent Text Reader

Abstract

To provide a pump inspection report generation method and a pump inspection report generation support system capable of automatically or semi-automatically creating an inspection report suitable for a user.SOLUTION: A pump inspection report generation method comprises the steps of: acquiring a fiberscope image of the inside of a pump; extracting a first captured image suitable for image processing; extracting a second captured image for visual inspection; and attaching the second captured image to a report on a deterioration condition generated from the first captured image.SELECTED DRAWING: Figure 2
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a pump inspection report generating method and a pump inspection report generating support system. [Background technology]

[0002] As pointed out in the Ministry of Land, Infrastructure, Transport and Tourism's draft manual for inspection, maintenance and renewal of river pumping equipment, pumping equipment at drainage pumping stations must be able to start reliably when needed in response to natural phenomena such as heavy rain, and must be able to perform sufficient drainage function without malfunction for the required period of time.During normal days, pumping is hardly operated, so the operating hours are short, but once flooding occurs, continuous operation is required, and the environment is different from that of normal regular use equipment, with high temperature, high humidity and low air pressure during operation, and long periods of shutdown in low temperatures during non-flooding periods.

[0003] On the other hand, pumping equipment at pumping stations is required to operate continuously once it has entered the operating period, and has the characteristic that inspection and maintenance must be carried out while the equipment is functioning. Furthermore, as a common feature of river pump equipment, the equipment is composed of many devices and equipment, and even if one of them breaks down, it will have some effect on the drainage function and in some cases lead to a malfunction, so the entire system must function reliably.

[0004] For example, a vertical shaft pump device, which is one example of a pump device, is configured to include a rotating shaft, an impeller attached to the rotating shaft, a submerged bearing that rotatably supports the rotating shaft, and a pump casing that houses the impeller and the submerged bearing. The vertical shaft pump device is operated with the casing, impeller, and submerged bearing submerged in water, and these components gradually corrode and wear over the course of operation.

[0005] For this reason, it is necessary to periodically inspect the vertical pump equipment to check the wear of the impeller and submersible bearings and the corrosion of the casing, and to repair or replace them as necessary. However, if the vertical pump equipment is dismantled and the vertical pump is pulled up using a crane or the like to inspect the parts to check the wear of the parts, this not only increases the equipment and labor costs for the inspection work, but also takes a long time to perform the inspection.

[0006] To address this problem, an endoscopic image is taken by manipulating a fiberscope from outside the pump casing to a specific consumable part inside the pump casing, as in Patent Document 1. Furthermore, Patent Document 1 describes a method for diagnosing the degree of deterioration by generating an evaluation value by image processing of the image taken by the fiberscope, and a learning mechanism using a neural network.

[0007] There is also known a pump maintenance inspection system that manages captured images for each imaging location and generates an inspection report to which the captured images are attached. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] JP 2019-20264 A Summary of the Invention [Problem to be solved by the invention]

[0009] The inventors referred to the above-mentioned conventional techniques, processed captured images, performed degradation diagnosis using a degradation diagnosis model, and generated an inspection report in which the images used in the image processing were attached together with the degradation diagnosis results. However, the inspection report did not turn out to be suitable in all cases. In particular, various maintenance and inspection guidelines require observation of the condition and progress of deteriorated parts, but the images that are suitable for visual inspection are different from the images that are suitable for image processing, and there were also cases where the same angle of view was more appropriate for monitoring the progress of deteriorated parts. In other words, previous pump maintenance and inspection systems did not take such perspectives into consideration, and therefore did not necessarily produce inspection reports that were suitable for users.

[0010] Therefore, the present invention provides a pump inspection report generating method and a pump inspection report generating support system that can automatically or semi-automatically create an inspection report that is suitable for a user. [Means for solving the problem]

[0011] In order to solve the above problems, the pump inspection report generating method of the present invention is characterized by comprising the steps of obtaining a fiberscope image of the inside of the pump, extracting a first captured image suitable for image processing, extracting a second captured image for visual inspection, and attaching the second captured image to a report on the deterioration status generated from the first captured image.

[0012] In addition, the pump inspection report generation support system of the present invention includes an image capture / diagnostic image data management unit that extracts a first image suitable for image processing from the acquired fiberscope images of the inside of the pump, and an aging deterioration comparison image search unit that extracts a second image for visual inspection from the acquired fiberscope images of the inside of the pump, and is characterized in that the second image is attached to a report on the deterioration status generated from the first image. Effect of the Invention

[0013] According to the present invention, it is possible to provide a pump inspection report generating method and a pump inspection report generating support system that can automatically or semi-automatically create an inspection report that is suitable for a user.

[0014] Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0015] [Figure 1] 1A and 1B are diagrams illustrating a schematic structure of a pump device and a diagnostic method. [Diagram 2] 1 is a schematic configuration diagram of a pump inspection report generation support system according to a first embodiment of the present invention. [Diagram 3] FIG. 3 is a diagram showing a data structure of a database shown in FIG. 2. [Figure 4] 4 is a flowchart showing an operation of the pump inspection report generation support system according to the first embodiment of the present invention. [Diagram 5] FIG. 1 is a conceptual diagram of an imaging diagnostic and report generation model. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. EXAMPLES

[0017] The inspection report generating system of this embodiment will be described below using a vertical pump as an example, although the pump is not limited to a vertical pump.

[0018] [Pump equipment and internal observation of pump] The pump in this embodiment is a river pump facility managed by the Ministry of Land, Infrastructure, Transport and Tourism, local governments, etc. These river pump facilities are equipped with large vertical pumps of several meters and several tons, and therefore cannot be easily disassembled for inspection. Therefore, it is necessary to check the deterioration state inside the pump using a fiberscope.

[0019] FIG. 1 is a diagram explaining the general structure of a pump device and the method of diagnostic work, showing the structure of a vertical pump and the state of internal observation using a fiberscope.

[0020] First, the structure of the vertical pump will be described. 1, the vertical pump is equipped with a suction bellmouth 21 located at the lower end, an impeller casing 22 connected to the upper end of the suction bellmouth 21, a suspension pipe 24 connected to the upper end of the impeller casing 22 for suspending the impeller casing 22 in a water tank 23, and a discharge bend pipe 25 connected to the upper end of the suspension pipe 24. The impeller casing 22, the suspension pipe 24, and the discharge bend pipe 25 constitute a "pump casing" that serves as a pumping pipeline.

[0021] An impeller 27 having a plurality of blades 26 is disposed within the impeller casing 22. A rotating shaft 28 is connected to this impeller 27. The rotating shaft 28 of this impeller 27 extends vertically through the discharge bent pipe 25, the suspension pipe 24, and the impeller casing 22, and is rotatably supported by a lower underwater bearing 29 within the impeller casing 22 and an upper underwater bearing 30 within the suspension pipe 24. The upper end of the rotating shaft 28 protrudes upward from the discharge bent pipe 25, and is connected to a drive source (not shown).

[0022] A holder 31 that holds the lower underwater bearing 29 is provided on the upper portion of the impeller 27. A plurality of guide vanes 32 are provided between the outer circumferential surface of the holder 31 and the inner surface of the impeller casing 22. The vertical pump constructed as described above extends downward through an insertion hole 34 formed in the pump installation floor 33 above the water tank 23 so that its suction bellmouth 21 and impeller casing 22 are submerged in the stored water in the water tank 23, and is fixed to the pump installation floor 33 via an installation base 35 provided on the upper end side of the suspension pipe 24. A valve 36 is provided on the discharge side of the discharge bend 25 to make the inside of the pump casing airtight, and can be opened and closed.

[0023] A compressed air supply pipe connection port 43 and a pressure gauge connection port 44 are provided on the upper side of the discharge bent pipe 25. The compressed air supply pipe connection port 43 and the pressure gauge connection port 44 are usually closed with lids. When inspecting the components inside the pump casing for damage, wear, etc., a compressed air supply pipe 45 is connected to the compressed air supply pipe connection port 43. The compressed air supply pipe 45 is linked to a compressor 46 that generates compressed air.

[0024] In preparation for observation, with the vertical pump stopped, first, the valve 36 on the discharge side of the discharge bend 25 is closed. Next, the compressed air supply pipe 45 connected to the compressor 46 is connected to the compressed air supply pipe connection port 43.

[0025] Next, the compressor 46 is driven to supply compressed air from the compressor 46 into the pump casing through the compressed air supply pipe 45, pushing down the water in the pump casing and replacing the submerged portion of the pump casing with compressed air. As a result, the water that has entered the pump is turbid and is excluded from the water containing debris and the like, so that an optical path for an image acquisition means such as a fiberscope can be secured for the observation locations such as the inner surface of the impeller casing 22, the multiple blades 26, impeller 27, rotating shaft 28, and lower submerged bearing 29 in the impeller casing 22. As a result, each part in the pump casing can be recognized reliably, improving observation performance.

[0026] Furthermore, water droplets, dust, etc. adhering to the observation target areas such as the inner surface of the impeller casing 22, the multiple blades 26 inside the impeller casing 22, the impeller 27, the rotating shaft 28, the lower submerged bearing 29, etc. are blown away by the compressed air, so the observation target areas can be observed in a clean state. Whether or not the submerged parts inside the pump casing have been replaced with compressed air can be confirmed by checking whether air bubbles flowing out from the suction port of the suction bell mouth 21 in the pump casing are generated on the water surface.

[0027] With the submerged portion inside the pump casing pressed down, the light source device, scanner, monitor, image recorder, etc., installed above ground, are then turned on, and while observing the image information from the optical lens at the tip of the fiber cable 37 of the fiberscope constituting the image acquisition means on the monitor, the tip of the fiber cable 37 is inserted into the pump casing through the pressure gauge connection port 44, and the tip of the fiber cable 37 is scanned by the scanner, thereby making it possible to observe the surface of the blades (impeller) 26 on the monitor. The observation data acquired by this fiber cable 37 can be stored in the image recorder as necessary.

[0028] In this way, when the vertical pump is stopped, the water remaining in the casing is pushed out by compressed air, filling the casing with air, and the internal parts can be observed with a fiberscope, making it possible to check the wear and damage of the parts inside the pump. This improves the accuracy of the assessment work for pump overhaul and increases the reliability of the maintenance management of pump equipment.

[0029] [Configuration of pump inspection report generation support system] FIG. 2 is a schematic diagram of the system configuration of the pump inspection report generation support system according to the present embodiment. 2, the pump inspection report generation support system 1 according to this embodiment is constructed on a virtual cloud or a physical server. An inspection data acquisition app on a mobile terminal such as a smartphone, a PC (Personal Computer), or a tablet held by a diagnostician (inspector) at the site and the pump inspection report generation support system 1 (cloud) transmit and receive data to and from each other via a WAN (Wide Area Network).

[0030] The pump inspection report generation support system 1 includes a captured image / diagnosis image data management unit 2, an image diagnosis output unit 3, a report output unit 4, an aged deterioration comparison image search unit 5, an AI output correction unit 6, a user management / access management unit 7, a deterioration diagnosis model 8, a report creation model 9, and a database 10, which can communicate (transfer data) with each other via an internal bus. Here, the captured image / diagnosis image data management unit 2, the image diagnosis output unit 3, the report output unit 4, the aged deterioration comparison image search unit 5, the AI ​​output correction unit 6, and the user management / access management unit 7 are realized by, for example, a processor such as a CPU (not shown), a ROM for storing various programs, a RAM for temporarily storing data in the calculation process, and a storage device such as an external storage device, and the processor such as a CPU reads and executes various programs stored in the ROM, and stores the calculation results, which are the execution results, in the RAM or the external storage device.

[0031] The user management / access management unit 7 performs user authentication and access authority management for logging in to the pump inspection report generation support system 1 in the cloud, making it possible to obtain data from the site. The captured image / diagnosis image data management unit 2 stores the captured image data uploaded from the site in the database 10. When storing the data, the pump to be inspected and the inspection implementation date or transmission date are stored in association (linked) with the captured image data and inspection record.

[0032] The image diagnosis output unit 3 inputs the captured image data stored in the database 10 by the captured image / diagnostic image data management unit 2 to an image diagnosis AI (not shown) incorporating a deterioration diagnosis model 8, and performs image diagnosis. The image diagnosis AI is, for example, configured with a CNN (Convolutional Neural Network) or the like. Specifically, it identifies the area (including dimensions) of the deteriorated part by edge detection in filter processing and binarization processing, masks the area, and estimates the cause of deterioration, and outputs the masked diagnostic image data and inspection record (diagnosis data) to the database 10. It is possible to identify corrosion, defects, or wear as the cause of deterioration by known techniques. Here, as shown in FIG. 3, the database 10 has a data structure that stores the captured image data 15, diagnostic image data 16, comparison standard / past image data 17, provisional report 18, and revised report (confirmed final report) 19 in separate areas. However, this is not necessarily limited to this configuration, and the captured image data 15, diagnostic image data 16, comparison standard / past image data 17, provisional report 18, and revised report (confirmed final report) 19 may be stored individually in hardware.

[0033] The aging deterioration comparison image search unit 5 extracts image data for display to be used in the report from the captured image data of the captured image / diagnosis image data management unit 2 or the diagnostic image data (masked and with the deteriorated area specified) by the image diagnosis output unit 3. In the extraction, comparison data (past captured image data, diagnostic image data, or reference image data of the pump to be inspected) is extracted from the related data of the image diagnosis output unit 3. By comparing at least one of the image capturing position, angle of view, and clarity between the current image data and the past image data (or reference image data), image data that is easy to visually compare for aging deterioration is automatically extracted. Alternatively, the image data for display is extracted based on the similarity between the current image data and the past image data or the reference image data. Past reports may be machine-learned. This makes it possible to automate the report creation work based on personal experience, thereby preventing it from becoming personal. In addition, the report becomes easy to use for customers. The past image data (or reference image data) is stored in the comparison standard / past image data 17, which is one area of ​​the database 10.

[0034] The report output unit 4 receives as input the diagnostic image data from the image diagnosis output unit 3, the inspection records, and the image data extracted by the aging deterioration comparison image search unit 5, and automatically generates a report using an AI incorporating a report automation model, and stores the report as a provisional report in a provisional report 18, which is one area of ​​the database 10. The AI ​​incorporating the report automation model is, for example, configured with a CNN (Convolutional Neural Network) or the like.

[0035] The AI ​​output correction unit 6 visually checks the provisional report from the report output unit 4 as necessary, and makes corrections as necessary. If corrections are made, the report is stored as a corrected report in the corrected report (confirmed final report) 19, which is one area of ​​the database 10.

[0036] [Operation of the pump inspection report generation support system] Next, the operation of the pump inspection report generation support system 1 in this embodiment will be explained using the flowchart shown in Figure 4, in relation to a situation in which a diagnostician (who in this embodiment also serves as an inspector) visits the site shown in Figure 2.

[0037] First, a diagnostician (who also serves as an inspector in this embodiment) visits the site. The diagnostician who visits the site launches an inspection application on a smartphone. The inspection data acquisition application prompts the inspector to take endoscopic images of the pump to be inspected and each inspection location. The inspector operates a fiberscope (endoscope) to take an endoscopic image of the inside of the pump. He starts up the on-site app, accesses the pump inspection report generation support system 1 (cloud) via the WAN, and starts up the pump maintenance and inspection system. The inspector enters a user ID and password to start up the pump inspection report generation support system 1. In other words, the user management and access management unit 7 performs user authentication and access authority management for logging in to the pump inspection report generation support system 1 in the cloud, making it possible to obtain data from the site.

[0038] In step S11, the captured image / diagnostic image data management unit 2 constituting the pump inspection report generation support system 1 stores the captured image data uploaded from the site in the database 10. Next, the captured endoscopic image and the inspector's inspection record are uploaded to the database 10 by the captured image / diagnostic image data management unit 2 via the WAN. The inspection record is an inspection memo prepared by the inspector to such an extent that an inspection report can be automatically generated by document analysis (document AI). Here, the document analysis itself can be performed using a known technology.

[0039] In step S12, the image diagnosis output unit 3 constituting the pump inspection report generation support system 1 inputs the captured image data stored in the database 10 by the captured image / diagnosis image data management unit 2 to an image diagnosis AI (not shown) incorporating the deterioration diagnosis model 8, and performs image diagnosis. As described above, specifically, the area (including dimensions) of the deteriorated part is identified by edge detection in filtering and binarization processing, the area is masked, and the cause of deterioration is estimated, and the masked diagnostic image data and inspection record (diagnosis data) are output to the database 10.

[0040] In step S13, the aging deterioration comparison image search unit 5, which is part of the pump inspection report generation support system 1, extracts display image data to be used in the report from the captured image data in the captured image / diagnostic image data management unit 2 or the diagnostic image data (after masking and identification of deteriorated areas) from the image diagnosis output unit 3.

[0041] In step S14, the report output unit 4 constituting the pump inspection report generation support system 1 inputs the inspection record and the image data extracted in step S13. In step S15, the report output unit 4, which is part of the pump inspection report generation support system 1, automatically generates a report using AI that has incorporated the report automation model, and stores the report as a provisional report in the provisional report 18, which is one area of ​​the database 10.

[0042] In step S16, the AI ​​output correction unit 6 constituting the pump inspection report generation support system 1 displays the interim report generated by the report output unit 4 to the inspector, prompting him / her to decide whether or not correction is necessary. If it is determined in step S16 that correction is necessary, the process proceeds to step S17, where the AI ​​output correction unit 6 creates a corrected report and ends the process. On the other hand, if it is determined in step S16 that correction is not necessary, the interim report created in step S15 is treated as a confirmed final report and the process ends. Although not shown, the inspector is prompted to access the interim report created in step S15 or the corrected report created in step S16, and confirms the interim report as the reporter. If correction is necessary, the data is corrected, and if correction is not necessary, only confirmation is performed. If correction is made, it is stored in the database as a corrected report. After the confirmed final report has been approved by the relevant parties, the reporter will download and print it out, submit it to the customer, and make a proposal for maintenance and repair of the pump.

[0043] Here, Fig. 5 is a conceptual diagram of the image diagnosis and report creation model. As shown in the "Image diagnosis" column in Fig. 5, when an image of the inside of a pump is input to the pump inspection report creation support system 1, the deterioration diagnosis model 8 (Fig. 2) uses artificial intelligence (AI) to mask defective areas such as corrosion, missing parts, and wear, and generates a masked image. To make it easier to view, the internal image 11 and the masked non-conformity image 12 are displayed side by side.

[0044] Furthermore, as shown in the "Report Automation" column in Figure 5, when the report creation model 9 inputs an image 12 in which masked non-conformance areas are output by the image analysis AI, and the inspection results 13, into the pump inspection report generation support system 1, a report document 14 is automatically output by a text analysis AI (not shown). Figure 5 shows a case in which the defective portion is masked (a mask is affixed), but in the case of corrosion, for example, it is sufficient to mask with a color different from that for the defective portion.

[0045] In this embodiment, the pump inspection report generation support system 1 is configured to have the AI ​​output correction unit 6, but this is not limited to this. In other words, the pump inspection report generation support system 1 may be configured not to have the AI ​​output correction unit 6, so that the report is automatically created.

[0046] As described above, according to this embodiment, it is possible to provide a pump inspection report generating method and a pump inspection report generating support system that can automatically or semi-automatically create an inspection report that is suitable for a user.

[0047] The present invention is not limited to the above-described embodiment, but includes various modified examples. For example, the above-described embodiment has been described in detail to easily explain the present invention, and the present invention is not necessarily limited to the embodiment having all of the described configurations. [Explanation of symbols]

[0048] 1...Pump inspection report generation support system, 2...Captured image / diagnosis image data management section, 3...Image diagnosis output section, 4...Report output section, 5...Aging deterioration comparison image search section, 6...AI output correction section, 7...User management / access management section, 8...Deterioration diagnosis model, 9...Report creation model, 10...Database, 11...Internal image, 12...Masked non-compliant image, 13...Inspection results, 14...Report materials, 15...Captured image data, 16...Diagnostic image data, 17...Comparison standard / past image data, 18...Tentative report Notification, 19...Amended report (confirmed final report), 21...Suction bell mouth, 22...Impeller casing, 23...Water tank, 24...Suspension pipe, 25...Discharge bend pipe, 26...Blade, 27...Impeller, 28...Rotating shaft, 29...Lower submerged bearing 29, 30...Upper submerged bearing, 31...Retainer, 32...Guide vane, 33...Pump installation floor, 34...Through hole, 35...Installation base, 36...Valve, 37...Fiber cable, 43...Compressed air supply pipe connection, 44...Pressure gauge connection, 45...Compressed air supply pipe, 46...Compressor

Claims

1. obtaining a fiberscope image of the interior of the pump; Extracting a first captured image suitable for image processing; extracting a second captured image for visual inspection; A pump inspection report generating method comprising the step of attaching the second captured image to a report on deterioration status generated from the first captured image.

2. 2. The pump inspection report generating method according to claim 1, further comprising: A pump inspection report generating method, characterized in that the extraction of the second captured image for visual inspection is performed based on similarity to a previously captured image or a reference image.

3. 2. The pump inspection report generating method according to claim 1, further comprising: A pump inspection report generating method, characterized in that the extraction of the second captured image for visual inspection is performed based on at least one of the imaging position, the angle of view, and the clarity.

4. an image / diagnosis image data management unit that extracts a first image suitable for image processing from among the acquired fiberscope images of the inside of the pump; an aging deterioration comparison image search unit that extracts a second captured image for visual inspection from among the acquired fiberscope images of the inside of the pump; A pump inspection report generation support system characterized in that the second captured image is attached to a report on the deterioration status generated from the first captured image.

5. The pump inspection report generation support system according to claim 4, The pump inspection report generation support system is characterized in that the aging comparison image search unit extracts the second captured image for visual inspection based on its similarity to a previously captured image or a reference image.

6. The pump inspection report generation support system according to claim 4, The pump inspection report generation support system is characterized in that the aging comparison image search unit extracts the second captured image for visual inspection based on at least one of an imaging position, an angle of view, and a sharpness.

Citation Information

Patent Citations

  • Deterioration diagnosis for LNG pump

    JP1990009969A

  • Pumping station checking system

    JP2003140735A

  • Inspection support processing server and inspection support system

    JP2014139724A

  • High-place facility inspection device and high-place piping inspection method

    JP2015023471A

  • Inspection support system, information terminal, inspection support method and program

    JP2017228249A