Evaluation of weld seams

Abstract

This invention provides a computer-based implementation method and system for evaluating non-destructive ultrasonic testing of plastic pipe welds. [Solution] In each of the following steps, The steps include: receiving an ultrasonic scanning data file using a server, The process involves a computing unit, including a neural network, analyzing ultrasound scanning data files based on predetermined criteria, and The calculation unit performs the step of evaluating the ultrasound scanning data file based on predetermined criteria. Computer implementation methods and systems, including those mentioned above.

Description

Technical Field

[0001] The present invention relates to a computer-implemented method for evaluating a non-destructive ultrasonic inspection of a plastic pipe weld, a system for evaluating a non-destructive ultrasonic inspection of a plastic pipe weld, and a method for training a neural network used in a method for evaluating a non-destructive ultrasonic inspection of a plastic pipe weld.

[0002] Today, the welding of plastic pipelines carried out at construction sites is inspected by ultrasonic inspection. Here, the ultrasonic inspection is carried out by an inspector who determines whether the weld meets the requirements and is qualified or does not meet the requirements and is unqualified based on the ultrasonic scanning. In this case, the inspector generally does not perform the evaluation on site himself / herself, but transfers the ultrasonic scanning to a test center, where the ultrasonic scanning is visually evaluated by a reviewer at this test center, and a determination as to whether the weld seam is qualified is sent to the inspector. The drawback here is that in most cases, the evaluation reaches the inspector several hours or even several days late. If the evaluation is not good, the welding and subsequent inspections must be carried out again or administered, and thus the site must be visited again, all of which are extremely time-consuming and costly.

[0003] European Patent Application Publication No. 3815884 discloses a method for inspecting a weld. In this method, the executed weld is recorded by a data processing device, and further, the defect location is detected and stored by ultrasonic waves. The drawback here is that the system does not directly evaluate the weld seam, but the inspector must send the ultrasonic inspection to a test center for review, or the test center must review and evaluate the data supplied to a web-based tool, which generally causes a delay of several hours to several days.

[0004] The object of the present invention is to propose a computer implementation method, a system connected to a computer, and a method for training a neural network that can provide real-time evaluation of ultrasonic scanning in plastic pipe welds, thereby significantly reducing the evaluation time for ultrasonic scanning.

[0005] The above problem is addressed by the present invention, which provides a computer-based method for evaluating non-destructive ultrasonic testing of plastic pipe welds, comprising the following steps: The steps include: receiving an ultrasonic scanning data file using a server, The process involves a computing unit, including a neural network, analyzing ultrasound scanning data files based on predetermined criteria, and The calculation unit performs the step of evaluating the ultrasound scanning data file based on predetermined criteria. This is resolved by computer implementation methods, including [specific methods].

[0006] The computer implementation method for evaluating non-destructive ultrasonic inspection of plastic pipe welds according to the present invention includes receiving ultrasonic scanning data files using a server. The server may include a computing unit and / or memory, which may also be configured as a GPU. The ultrasonic scanning data files can be transmitted directly from the ultrasonic scanner to the server, but they can also be transmitted to the server by another method, for example, via a USB stick or Bluetooth, in which case it is advantageous that the server is web-based and therefore accessible via the internet, and that the ultrasonic scanning data files can be transmitted to and received by the server. Here, the ultrasonic scanning data files are mappings of ultrasonic scanning of the plastic pipe weld, which is the weld to be inspected. Preferably, the ultrasonic scanning data files show the weld extending around the entire circumference, i.e., over 360°. The computer implementation method according to the present invention includes analyzing the ultrasonic scanning data files based on predetermined criteria using a computing unit. Preferably, the computing unit is a component of the server, but may be configured autonomously. The computing unit includes a neural network or learning algorithm, preferably a convolutional neural network (CNN), for analyzing and evaluating the ultrasonic scanning data files. Based on an evaluation of the ultrasound scanning data file using predetermined criteria, the final evaluation is preferably made as "pass" or "fail."

[0007] Preferably, the ultrasonic scanning data files are stored in a server or an alternative memory module.

[0008] Preferably, predetermined criteria and sub-criteria are stored in a database on the server or an alternative memory module, and are continuously expanded by new ultrasonic scanning data files and their evaluations. However, the database may be stored in a location separate from the server.

[0009] It is advantageous to divide the criteria into multiple inspection groups. This allows for efficient inspection and rapid evaluation by the calculation unit, resulting in substandard ultrasonic scanning data files not being used at all for inspection of weld quality, and the evaluation being stopped early.

[0010] Preferably, the inspection group defines the quality of the ultrasound scanning data file. It is advantageous for the computing unit to first evaluate the quality group of the ultrasound scanning data file using a neural network. If the ultrasound scanning data file is evaluated as "failure," the inspection is interrupted, and a message indicating that the ultrasound scanning data file was "failure" is output. This message is preferably displayed directly to the on-site inspector via a display device and, if necessary, transmitted to the test center via a web-based platform. Based on this, the ultrasound inspection is performed again, the ultrasound scanning data file is transmitted to the server again, and then analyzed and evaluated by the computing unit.

[0011] It is advantageous that the quality of ultrasonic scanning data files be evaluated based on criteria for the extent of undulations shown in the ultrasonic scanning data files, particularly for weld seams and the outer surface of pipes. Preferably, sub-criteria include whether the undulations extend continuously without interruption, whether the undulations are uniformly formed, and whether scratches are present in the undulations. Of course, this list is not exhaustive, and other sub-criteria may be added.

[0012] Preferably, the quality of the ultrasonic scanning data file is evaluated based on criteria for the extent of the wavy pattern shown in the ultrasonic scanning data file, particularly for the weld seam and the inner surface of the pipe. Preferably, sub-criteria include whether the wavy pattern extends continuously without interruption, whether the wavy pattern is uniformly formed, and whether scratches are present in the wavy pattern. Of course, this list is not exhaustive, and other sub-criteria may be added.

[0013] It is advantageous for the quality of ultrasonic scanning data files to be evaluated based particularly on the extent of wall thickness extension within the ultrasonic scanning data files of weld seams and pipes. Preferably, a sub-criterion exists below this: whether the wall thickness extends across the ultrasonic scanning data file or has the same width. Furthermore, as another sub-criterion, irregularity in the ultrasonic scanning with respect to wall thickness may also be examined and evaluated. Of course, this enumeration is not limiting, and other sub-criteria may be added.

[0014] It is advantageous for the quality of ultrasonic scanning data files to be evaluated based particularly on the presence of heated wires within the ultrasonic scanning data file. This criterion applies when ultrasonic inspection of welds using electric welding sleeves is performed. Preferably, both the number of identifiable heated wires or windings and the thickness or diameter of the heated wires present are inspected. If the calculation unit cannot identify these, the ultrasonic scanning data file is classified as insufficient and must be newly created.

[0015] It is advantageous for the quality of ultrasound scanning data files to be judged based, in particular, on whether the cold zone is identifiable within the ultrasound scanning data file. Furthermore, it is advantageous for the length of the cold zone to be recorded together as a quality criterion for ultrasound scanning.

[0016] It is advantageous for the quality of ultrasonic scanning to be evaluated based on the criteria of the ultrasonic scanning settings performed, particularly for weld seams and pipes. Preferably, these criteria include, below that, the speed of creating ultrasonic scanning data files and zero line identification, or the calibration of the ultrasonic scanner for accurately detecting ultrasonic scanning data files. Of course, this enumeration is not limiting, and other sub-criteria may be added.

[0017] It has been found that it is advantageous for the evaluation of individual criteria to be based on a percentage subtraction from a 100% base. That is, for example, in a quality inspection group for ultrasonic scanning data files, if the outer surface criterion is inspected and the calculation unit identifies that the extension of the wavy portion is inconsistent, an evaluation of, for example, 2% is output based on the defect location in the wavy portion, and this is then subtracted from the current optimal base value of 100%.

[0018] It is advantageous if the evaluation of each of these criteria or sub-criteria is performed by a computation unit.

[0019] If the quality inspection of the ultrasound data file fails, it has been found to be advantageous for the computing unit to make suggestions for achieving sufficient quality in the ultrasound scanning data file, such as suggesting cleaning at a test center.

[0020] Preferably, after inspection and evaluation of the criteria for the quality of the ultrasonic scanning data file, inspection and evaluation of the criteria for the quality of the welded part are performed using the computer implementation method according to the present invention.

[0021] It has been found that a preferred embodiment is one in which the quality of the weld is inspected and evaluated based on at least one of the following criteria: abnormalities, defects, or unexpected non-uniform structures in the ultrasonically scanned material, or the absence of a heating wire. Of course, this enumeration is not limiting, and other sub-criteria may be added.

[0022] Preferably, the ultrasound examination is performed using the Time of Flight Diffraction (TOFD) method or Phase Array Ultrasound (PAUT).

[0023] Preferably, butt welds, electric welds using electric welding sleeves, and other plastic pipe welds are evaluated. A predetermined criterion is set for the plastic pipe weld to be analyzed, such as applying heated wire inspection only to the areas where heated wire has been applied.

[0024] The above problems are also solved by a system for evaluating non-destructive ultrasonic inspection of plastic pipe welds according to the present invention, comprising: an ultrasonic scanner that performs ultrasonic scanning of the plastic pipe weld and creates an ultrasonic scanning data file; a server that receives the ultrasonic scanning data file; and a computing unit that analyzes the ultrasonic scanning data file and evaluates the ultrasonic scanning data file based on predetermined criteria, wherein the computing unit comprises a neural network and an output unit that displays the evaluation. Preferably, the final evaluation is made as "pass" or "fail". The ultrasonic image of the weld detected by the ultrasonic scanner is transmitted to or received by the server. Preferably, a web-based platform to which the ultrasonic scanning data file is transmitted is used, but the ultrasonic scanning data file may be transmitted to the server by another method. It is advantageous for the server to have a computing unit and memory, but each unit can also be arranged separately from each other. The computing unit analyzes and evaluates the ultrasonic scanning data file based on the criteria already enumerated above for computer implementation methods. The computing unit analyzes and evaluates the criteria using a neural network. The evaluation is notified to the inspector on-site by the output unit.

[0025] It is advantageous for the server to have a web-based interface. This allows on-site inspectors to access ultrasonic scanning data files on-site, just as they would at a remote test center.

[0026] The above problem is further solved by a method for training a neural network used in a method for evaluating a non-destructive ultrasonic inspection of a plastic pipe welded part according to the present invention, the method comprising the following steps, namely, · creating a simulated ultrasonic scanning data file using the neural network of the computing unit based on known ultrasonic scanning data files for various materials; · evaluating the ultrasonic scanning data file based on a criterion; · continuously expanding the neural network based on the simulated ultrasonic scanning data file and is also solved by a method including.

[0027] Preferably, the server or the computing unit has a simulation unit capable of creating further ultrasonic data files based on known materials using the neural network, thereby expanding the diversity of the ultrasonic data files for utilization of the evaluation and expanding the database. By doing so, mainly the intermediate region can be reduced, thereby ensuring that no irrelevant evaluations are created.

[0028] A method for training a neural network used in a method for evaluating a non-destructive ultrasonic inspection of a plastic pipe welded part has, as a predetermined criterion, at least one of the following criteria, namely, the state of extension of the undulations shown in the ultrasonic scanning data file for the welded seam and the outer or inner surface of the pipe, the state of extension of the wall thickness in the ultrasonic scanning data file of the welded seam and the pipe, an abnormality, a defect or an unexpected non-uniform structure in the ultrasonic scanning data file, or the absence of a heating wire. Of course, further different criteria for training the neural network can also be used, and the criteria described above in relation to the method can also be used for training the neural network.

[0029] It is advantageous if the training or expansion of the database is performed periodically or automatically each time a new evaluation of the ultrasonic scan data file is performed.

[0030] It has been found to be advantageous that the neural network is a convolutional neural network (CNN). The neural network or learning algorithm is trained to obtain the ultrasonic scan data file of the welded joint as an input and calculate a comparable criterion for the existing ultrasonic scan data file therefrom. Preferably, adjustment values, material data or welding parameters that may also exist within the neural network can also be used for assistance.

[0031] <0,000102>It is pointed out that all the constituent means can be freely combined with each other, and in order to avoid repetition, the features of the computer-implemented method and system are automatically related to the training method and vice versa.

[0032] Embodiments of the present invention will be described with reference to the drawings. However, the present invention is not limited to only these embodiments. The following is shown in the drawings.

Brief Description of the Drawings

[0033] [Figure 1] It is a flowchart showing a computer-implemented method according to the present invention. [Figure 2] It is a diagram showing an ultrasonic data file including marked defect positions.

[0034] Figure 1 shows a flowchart of a computer implementation method for evaluating non-destructive ultrasonic inspection of plastic pipe welds according to the present invention. The ultrasonic data file USSD is received using a server in IN USSD. Subsequently, a calculation unit performs analysis and evaluation against predetermined criteria. Here, the analysis and evaluation 1QUSSD is associated with the quality of the ultrasonic data file USSD to ensure that further evaluation of the weld is performed using only ultrasonic scanning data files that meet the requirements. Thus, the criteria for 1QUSSD of the first inspection group are related to the quality of the scan. Preferably, the criteria for analysis and evaluation of the ultrasonic scanning data file are the extension state of the corrugation shown in the ultrasonic scanning data file for the weld seam and the outer and inner surfaces of the pipe, and / or the extension state of the wall thickness in the ultrasonic scanning data file for the weld seam and pipe. Below this, sub-criteria are included, for example, whether the corrugation extends continuously without interruption, whether the corrugation is uniformly formed, or whether scratches are observed in the corrugation. Of course, this enumeration is not limiting, and other sub-criteria can be added, or only individual criteria from those enumerated can be selected. As a sub-criterion for the extent of wall thickness in ultrasonic scanning data files of weld seams and pipes, for example, whether the wall thickness extends consistently across the ultrasonic scanning data file can be analyzed and evaluated. As yet another sub-criterion, irregularities in wall thickness within the ultrasonic scanning data file can also be inspected and evaluated.

[0035] It has been found to be advantageous when the evaluation of individual criteria is based on a percentage subtraction from a 100% base. For example, in a quality inspection group for an ultrasonic scanning data file (USSD), if the outer surface criterion is inspected and the calculation unit identifies that the extension of the wavy portion is inconsistent, an evaluation of, for example, 2% is output based on the defect location in the wavy portion, and this is then subtracted from the output optimal base value of 100%.

[0036] The following is an assessment of the quality of the ultrasound scanning data files for the first group of tests: [Table 1-1] [Table 1-2]

[0037] Here, the evaluation reveals that the quality of the ultrasonic scanning data file from 1QUSSD is insufficient. Therefore, in the flowchart in Figure 1, the WELD REP step is subsequently performed to replace the weld, and the weld is removed and re-fabricated. Next, a new ultrasonic inspection CR USSD is performed, and the created ultrasonic scanning data file is transmitted to or received by the server. After that, the ultrasonic scanning data file runs through the first inspection 1QUSSD again.

[0038] [Table 2-1] [Table 2-2]

[0039] Here, we can see that 100% has been achieved, and therefore, the first inspection regarding the quality of the ultrasonic scanning data file has been passed.

[0040] Subsequently, a second inspection, namely the weld inspection (2QWELD), is performed. Here, too, individual criteria are preferably evaluated in percentage terms. For example, if an abnormality occurs, it is evaluated as a corresponding percentage value based on its size, shape, and quantity, and this percentage value allows for welding calculations.

[0041] [Table 3]

[0042] The example above shows an evaluation that the weld failed. Following such an evaluation, the ultrasonic scanning data file (USSD) is transmitted to the test center or provided to a web-based platform that allows the test center to perform a visual inspection. In this case, the test center visually evaluates the ultrasonic scanning data file with a human reviewer. Thus, welds that failed the neural network evaluation are now evaluated by the reviewer, who makes a final determination as to whether the weld meets the requirements and passes, or whether it does not and must be removed, and the process restarts.

[0043] If the weld is judged "yes" by the test center, the output unit OUTPUT will output that the weld has passed, either via a digital display or a printout.

[0044] If the system equipped with the neural network according to the present invention evaluates the weld as 100%, evaluation by the test center is no longer necessary, and this evaluation is directly supplied to the output unit OUTPUT along with the "pass" rating. Please refer to the following table for details.

[0045] [Table 4]

[0046] Of course, each evaluation can be adapted individually. Reaching 100% is not a requirement for passing; a value of 90% or even a more specifically set value may be required.

[0047] Figure 2 shows an ultrasound scanning data file (USSD) where, as an example, both inspections were performed, but both the first and second inspections failed. However, normally, after the first inspection for scan quality, if this inspection fails, the inspection is stopped. Except for point 5, where the surface is inconsistent, the consistent and uniform extension of the wavy portion on the inner surface 1 and outer surface 2 is clearly visible. The wall thickness extends consistently. Thus, although the standard and sub-standards here are evaluated at specific percentages, the first inspection for the quality of the ultrasound scanning data file fails, not 100%. The ultrasound scanning data file (USSD) also shows several defects in the tissue surrounded by frames, and the computer implementation method preferably characterizes these locations by fencing, as can be seen in Figure 2. The defects in the ultrasonic scanning data file USSD were evaluated using the computer implementation method according to the present invention. Although the values ​​at that time are not shown here, the final value of the second inspection regarding the quality of the weld did not reach 100%, so the ultrasonic scanning data file USSD was provided to the test center for visual inspection by the examiner. As shown in these test images in Figure 2, if there are many defects, the examiner also evaluates the weld as unacceptable, and the inspector instructs that the weld be removed and re-inspected.

[0048] It is advantageous to evaluate the quality of ultrasonic scanning based on the settings of the ultrasonic scanning performed on weld seams and pipes. Preferably, sub-criteria include, under these, the speed of creation of ultrasonic scanning data files and identification of the zero line, or the calibration of the ultrasonic scanner for accurate detection of ultrasonic scanning data files. Of course, this enumeration is not limiting, and other sub-criteria may be added. [Explanation of symbols]

[0049] USSD Ultrasound Data File IN USSD receives ultrasound scanning data files. 1QUSSD First inspection group, quality of ultrasound scanning data files 2QWELD Second inspection group, quality of welds WELD REP: Replace welded joints. Create CR USSD ultrasound scanning data files Visual evaluation by the VISUAL Test Center OUTPUT Evaluation output 1. Wavy part of the inner surface 2. Wavy portion of the outer surface 3 Wall thickness 4. Defects 5. The wall thickness is not consistent throughout.

Claims

1. A computer implementation method for evaluating non-destructive ultrasonic testing of plastic pipe welds, comprising the following steps: The steps include receiving an ultrasonic scanning data file (USDD) using a server, The process involves a computing unit including a neural network to analyze the ultrasound scanning data file (USSD) based on predetermined criteria, and The calculation unit performs the steps of evaluating the ultrasonic scanning data file (USD) based on the predetermined criteria and including, Computer implementation method.

2. The aforementioned criteria are divided into multiple test groups (1 QUSSD, 2 QWELD). The computer implementation method according to claim 1.

3. The aforementioned inspection group defines the quality of the ultrasound scanning data file (1QUSSD). The computer implementation method according to claim 2.

4. The evaluation of the quality of the aforementioned ultrasonic scanning data file (USSD) includes criteria for the extent of wavy areas shown in the ultrasonic scanning data file (USSD) for weld seams and the inner and / or outer surfaces of the pipe. The computer implementation method according to claim 3.

5. The evaluation of the quality of the aforementioned ultrasonic scanning data file (USSD) includes criteria for the extension of the wall thickness of the weld seam and pipe within the ultrasonic scanning data file (USSD). The computer implementation method according to claim 3.

6. The evaluation of the quality of the aforementioned ultrasonic scanning data file (USSD) includes a criterion for the presence of heating wires within the ultrasonic scanning data file (USSD). The computer implementation method according to claim 3.

7. One inspection group defines the quality of the weld (2QWELD). The computer implementation method according to claim 2.

8. The evaluation of the quality of the welded joint includes at least one of the following criteria: namely, the absence of abnormalities, defects, or unexpected non-uniform structures in the ultrasonic scanning data file, and the absence of heated wire. The computer implementation method according to claim 1.

9. The ultrasound examination used to create the ultrasound scan data file (USSD) is performed and created using either Time of Flight Diffraction (TOFD) or Phase Array Ultrasound (PAUT). The computer implementation method according to any one of claims 1 to 8.

10. A system for evaluating non-destructive ultrasonic testing of plastic pipe welds, An ultrasonic scanner that performs ultrasonic scanning of plastic pipe welds and creates an ultrasonic scanning data file (USDD), A server that receives the ultrasound scanning data file (USDD), A computing unit that analyzes the ultrasound scanning data file (USD) and evaluates the ultrasound scanning data file based on predetermined criteria. Includes, The aforementioned computing unit includes a neural network and an output unit (OUTPUT) that displays the evaluation. system.

11. The server has a web-based interface. The system according to claim 10.

12. Preferably, a method for training a neural network used in a method for evaluating non-destructive ultrasonic testing of plastic pipe welds according to any one of claims 1 to 9, comprising the following steps: The steps include creating simulated ultrasonic scanning data files (USSDs) using a neural network in a computing unit, based on known ultrasonic scanning data files (USSDs) for various materials, and A step of evaluating the ultrasound scanning data files (1 USDD, 2 QWELD) based on predetermined criteria, The steps include: continuously expanding the neural network based on the simulated ultrasound scanning data file (USD); and including, method.

13. At least one of the predetermined criteria is that the extension of wavy areas shown in the ultrasonic scanning data file for the weld seam and the outer or inner surface of the pipe, the extension of the wall thickness of the weld seam and pipe in the ultrasonic scanning data file, no abnormalities, defects or unexpected non-uniform structures, or heated wires are found in the ultrasonic scanning data file. The method according to claim 12.

14. The method according to claim 12, wherein the neural network is a convolutional neural network (CNN).

Images