Weld assessment

Abstract

A computer-implemented method and system for evaluating non-destructive ultrasonic inspection of plastic pipe welds, comprising the steps of: receiving an ultrasonic scan file by means of a server, analyzing the ultrasonic scan file by a computing unit on the basis of predefined criteria, and evaluating the ultrasound scan file by means of the computing unit according to predefined criteria.

Description

Technical Field

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

[0002] Currently, ultrasonic inspection is used to check welds on plastic pipes installed on construction sites. This inspection is performed by an inspector who assesses the weld's compliance with requirements based on ultrasonic scans, determining whether it passes or fails. The inspector typically does not perform the assessment on-site but instead forwards the ultrasonic scans to an inspection agency. The agency then evaluates the ultrasonic scans visually and sends an assessment indicating whether the weld is acceptable or unacceptable. The disadvantages of this approach are that the assessment is often delayed by hours or even days, and in cases of a negative assessment, the inspector must perform or request re-welding and subsequent inspection, necessitating a return trip to the construction site – all of which are time-consuming and costly.

[0003] EP 3815884 A1 discloses a method for inspecting welds, wherein the weld being performed is recorded by means of a data processing device, and defects are detected and stored by means of ultrasound.

[0004] The drawback here is that the system does not directly evaluate the weld; instead, the inspector must send the ultrasonic test results to the inspection agency for evaluation or instruct the inspection agency to inspect and evaluate the data input into the web-based tool, which typically results in delays of several hours or days. Summary of the Invention

[0005] The purpose of this invention is to provide a computer-implemented method, a system thereto, and a method for training a neural network that provides real-time evaluation of ultrasonic scans for welding plastic pipes, thereby significantly reducing the evaluation time of ultrasonic scans.

[0006] According to the present invention, this objective is achieved by a computer-implemented method for evaluating non-destructive ultrasonic testing of welded plastic pipes, the method comprising the following steps:

[0007] • Receive ultrasound scan files using a server.

[0008] • Analyzes ultrasound scan files based on predefined standards using computational units, where the computational units include neural networks.

[0009] • The ultrasound scan file is evaluated according to predefined criteria by a calculation unit.

[0010] A computer-implemented method according to the invention for non-destructive ultrasonic inspection of plastic pipe welds includes receiving ultrasonic scan files via a server. The server may include a computing unit and / or memory, or may be configured as a GPU. The ultrasonic scan files can be transferred directly from the ultrasonic scanner to the server, or sent or transmitted to the server in other ways, such as via a USB stick or via Bluetooth. Advantageously, the server is network-based and therefore accessible via the Internet, and the ultrasonic scan files are transmitted and can be received by the server. The ultrasonic scan files are images of the plastic pipe weld, i.e., the weld to be inspected. The ultrasonic scan files preferably show a complete circumferential weld extending 360°. The computer-implemented method according to the invention includes: the computing unit analyzing the ultrasonic scan files according to predefined criteria. The computing unit is preferably part of the server, but the computing unit can also be configured autonomously. The computing unit includes a neural network or learning algorithm, preferably a convolutional neural network (CNN), used to analyze and evaluate the ultrasonic scan files. The ultrasonic scan files are evaluated according to predefined criteria, and a final evaluation is performed, preferably "pass" or "fail".

[0011] Preferably, the ultrasound scan files are stored on a server or an alternative storage module.

[0012] Preferably, predefined standards and sub-standards are stored in a database on a server or alternative storage module and are continuously expanded with new ultrasound scan files and their evaluations. The database may also be stored in a location different from the server.

[0013] The advantage is that the standards are divided into inspection groups. This allows for efficient inspection and rapid evaluation through computational units, enabling the evaluation to be stopped early for poor-quality ultrasonic scan files, even without checking the weld quality.

[0014] Preferably, the inspection group defines the quality of the ultrasound scan file. Advantageously, the computing unit preferably uses a neural network to first evaluate the quality of the ultrasound scan file within the inspection group. If the ultrasound scan file is evaluated to the point that it is classified as "failed," the inspection is terminated, and a message is output: the ultrasound scan file "failed." This message is preferably displayed directly to the on-site inspector via a display device and, if necessary, forwarded to the inspection agency via a network-based platform. Based on this, another ultrasound inspection is performed, and the new ultrasound scan file is transmitted to the server so that it can then be analyzed and evaluated by the computing unit.

[0015] Advantageously, the quality of ultrasonic scan files is evaluated, particularly based on criteria regarding the wave trajectory of the weld and the outer surface of the pipe as shown in the ultrasonic scan files. Preferably, the criteria include sub-criteria such as whether the wave trajectory is continuous and uninterrupted, whether the wave structure is consistent, and whether burrs are visible in the wave. Of course, this list is not final and other sub-criteria may be added.

[0016] Preferably, the quality of the ultrasonic scan file is evaluated, particularly based on criteria regarding the wave trajectory of the weld and the inner surface of the pipe as shown in the ultrasonic scan file. Preferably, the criteria include the following sub-criteria: whether the wave trajectory is continuous and uninterrupted, whether the wave structure is consistent, and whether burrs are visible in the wave. Of course, this list is not final and other sub-criteria may be added.

[0017] Advantageously, the quality of ultrasonic scan files is evaluated, particularly based on the standard of wall thickness orientation in the ultrasonic scan files of welds and pipes. Preferably, the standard includes sub-standards such as whether the wall thickness has a constant orientation in the ultrasonic scan file, or whether the width is uniform. Furthermore, as another sub-standard, the presence of wall thickness irregularities in the ultrasonic scan can be checked and evaluated. Of course, this list is not final and other sub-standards can be added.

[0018] Advantageously, the quality assessment of ultrasonic scan files includes criteria based on the presence of heating wires within the ultrasonic scan files. These criteria are applied when ultrasonically inspecting welds using an electrofusion sleeve. Preferably, the number of identifiable heating wires or coils, as well as the thickness or diameter of the present heating wires, are also checked. If the calculation unit cannot identify these conditions, the ultrasonic scan file will be classified as unacceptable and must be recreated.

[0019] Advantages include: evaluating the quality of ultrasound scan files, particularly based on whether cold zones are identifiable in the scan files. Furthermore, it is advantageous to use the length of the cold zone as a quality standard for ultrasound scans.

[0020] Advantageously, the quality of ultrasonic scan files is evaluated, particularly based on the standards set for the ultrasonic scans performed on the welds and pipes. Preferably, the standards include sub-standards such as the speed at which ultrasonic scan files are created, and the identification of the zero line or the calibration of the ultrasonic scanner for proper detection of the ultrasonic scan files. Of course, this list is not final and other sub-standards may be added.

[0021] It has proven advantageous to evaluate each criterion based on a percentage deduction from a 100% baseline. That is, for example, in an inspection group checking the quality of ultrasonic scan documents, if the outer surface criterion is examined, and the calculation unit identifies a discontinuity in the wave direction, then an evaluation of, for example, 2% is output based on the defect in the wave, and that 2% is then subtracted from the initial optimal baseline value of 100%.

[0022] The advantage is that such an evaluation can be performed for each standard or sub-standard by means of a computational unit.

[0023] It has been proven advantageous that, in the event that the ultrasound document quality check fails, the calculation unit outputs suggestions for achieving sufficient quality in the ultrasound scan document, such as cleaning the inspection mechanism.

[0024] Preferably, after checking and evaluating the quality of the ultrasonic scan documents according to the standards of the inspection group, the inspection and evaluation of the welding quality is carried out by means of a computer-implemented method according to the invention.

[0025] A preferred implementation is to inspect and evaluate the quality of the weld according to at least one of the following criteria: anomalies, defects or inconsistencies in the ultrasonic scan file, unexpected structures, or failure to identify the heating wire. Obviously, this list is not final and other criteria may be added.

[0026] Preferably, the ultrasound examination used to create the ultrasound scan file is performed by means of time-of-flight diffraction (TOFD) or phased array ultrasound examination (PAUT).

[0027] Preferably, the evaluation includes butt welds, welds with welded sockets, and other plastic pipe welds. Based on the plastic pipe weld to be analyzed, predefined standards are confirmed: for example, heating wire inspection is only applied at locations where heating wires are used.

[0028] According to the invention, this objective is also achieved by a system according to the invention for non-destructive ultrasonic inspection of welded plastic pipes, the system comprising: an ultrasonic scanner for performing ultrasonic scanning on the welded plastic pipe and creating an ultrasonic scan file; a server for receiving the ultrasonic scan file; a computing unit for analyzing the ultrasonic scan file and evaluating the ultrasonic scan file based on predefined criteria, wherein the computing unit includes a neural network; and an output unit for displaying the evaluation. Preferably, the final evaluation is performed as "pass" or "fail". The ultrasonic images of the weld detected by the ultrasonic scanner are transmitted to or received by the server. For this purpose, a network-based platform is preferably used, and the ultrasonic scan file is sent to the platform, although it can also be transmitted to the server in other ways. Advantageously, the server has a computing unit and memory; however, the units can also be arranged separately from each other. The computer unit analyzes and evaluates the ultrasonic scan file according to the criteria listed above regarding the computer-implemented method. The computer unit analyzes and evaluates these criteria by means of a neural network. The evaluation results are communicated to the inspector in the field on the output unit.

[0029] The advantage is that the server has a network-based interface. This allows inspectors to access ultrasound scan files both on-site and from remote inspection facilities.

[0030] According to the present invention, this objective is also achieved by: a method for training a neural network according to the present invention, the neural network being used in a method for non-destructive ultrasonic inspection of plastic pipe welding, the method for training the neural network comprising the following steps:

[0031] • Based on known ultrasonic scan files of different materials, simulated ultrasonic scan files are created using a neural network of a computing unit.

[0032] • Evaluate ultrasound scan files according to standards.

[0033] • A continuously expanding neural network based on simulated ultrasound scan files.

[0034] Preferably, the server or computing unit has a simulation unit that can create additional ultrasound files based on known materials using a neural network, thereby increasing the diversity of ultrasound files used for evaluation and expanding the database. This primarily reduces the intermediate range, thus ensuring that inconsistent evaluations are not created.

[0035] Advantageously, the method for training the neural network used in the non-destructive ultrasonic inspection of plastic pipe welds has at least one of the following criteria as predefined standards: the wave orientation of the weld and the outer or inner surface of the pipe shown in the ultrasonic scan file; the orientation of the wall thickness in the ultrasonic scan file of the weld and the pipe; abnormalities, defects, or inconsistent, unexpected structures in the ultrasonic scan file; or the absence of a heating wire. Clearly, other criteria can also be used to train the neural network, as can the criteria mentioned earlier regarding the method.

[0036] The advantage is that training or database expansion can be performed periodically, or this can be done automatically each time ultrasound scan files are re-evaluated.

[0037] It is advantageous that the neural network is a convolutional neural network (CNN). The neural network or learning algorithm is trained to take ultrasonic scan files of the weld as input in order to derive similar standards to existing ultrasonic scan files. Preferably, settings, material data, or welding parameters that can also exist in the neural network can also be used for support.

[0038] All design options can be freely combined with each other, and to avoid duplication, the characteristics of computer-implemented methods and systems are automatically related to the methods used for training, and vice versa. Attached Figure Description

[0039] An embodiment of the present invention is described with reference to the accompanying drawings, but the invention is not limited to this embodiment. Wherein:

[0040] Figure 1 A flowchart illustrating a computer-implemented method according to the present invention is provided, and

[0041] Figure 2 An ultrasound file showing a marked defect is shown. Detailed Implementation

[0042] Figure 1 A flowchart illustrating a computer-implemented method for evaluating non-destructive ultrasonic inspection of plastic pipe welds according to the present invention is shown. Ultrasonic scan files (USSDs) are received via a server. Then, analysis and evaluation of predefined criteria are performed using a computing unit. The analysis and evaluation of USSDs (1QUSSDs) involves the quality of the USSDs to ensure that further evaluation of the weld is performed only with compliant ultrasonic scan files. Therefore, the criteria for the USSDs of the first inspection group relate to scan quality. The criteria used for analyzing and evaluating ultrasonic scan files are preferably the wave orientations of the weld and the inner and outer surfaces of the pipe as shown in the ultrasonic scan files, and / or the orientation of the wall thickness in the ultrasonic scan files of the weld and the pipe. For example, criteria may include sub-criteria such as whether the waves have a continuous, uninterrupted orientation, whether the waves are uniformly structured, or whether burrs are visible in the waves. Of course, this list is not final and other sub-criteria may be added, or only individual criteria from the listed criteria may be selected.

[0043] For example, it can be used as a sub-criterion for analyzing and evaluating the orientation of wall thickness in ultrasonic scan files of welds and pipes: whether the wall thickness is constant in the ultrasonic scan files. Furthermore, as another sub-criterion, irregularities in wall thickness during ultrasonic scans can be checked and evaluated.

[0044] It has proven advantageous to evaluate each criterion based on a percentage deduction from a 100% baseline. That is, for example, in a quality inspection group, the standard for inspecting the outer surface via a USSD ultrasonic scan document, and the computer unit identifies that the wave direction is discontinuous, then outputs an evaluation of, for example, 2% based on the defect in the wave, and then subtracts the 2% evaluation from the optimal baseline value based on 100%.

[0045] For the first inspection group regarding the quality of ultrasound scan documents, this assessment can be performed as follows:

[0046]

[0047]

[0048] As can be seen here: the assessment concludes that, according to 1QUSSD, the quality of the ultrasound scan files does not meet the requirements. Therefore, in Figure 1 As seen in the flowchart: the next step is the WELD REP (Weld Replacement) step, which removes and reconstructs the welded area. Subsequently, the CRUSSD (Ultrasonic Inspection System) is re-executed, and the created ultrasonic scan file is either retransmitted to the server or received by the server. The ultrasonic scan file will then be re-observed through the first inspection 1QUSSD.

[0049]

[0050]

[0051] As can be seen here: 100% has been achieved, and thus the first check on the quality of the ultrasound scan document has been passed.

[0052] The second inspection, 2QWELD welding, is then performed. Here, the various criteria are also preferably evaluated using percentages. For example, if an anomaly is found, it is evaluated as a corresponding percentage based on its size, shape, and quantity, which then allows for the verification of the weld.

[0053]

[0054]

[0055] The example above illustrates the following assessment, according to which the weld fails. Following this assessment, the ultrasonic scan file (USSD) is transmitted to an inspection agency, or provided to an inspection agency on a network-based platform, for visual inspection. The inspection agency then performs a visual assessment of the ultrasonic scan file by a human inspector. Thus, the weld, which was assessed as failing by the neural network, is now evaluated by the inspector, and this evaluation ultimately determines whether the weld is compliant and passes, or non-compliant and must be removed, and the process restarts.

[0056] If the inspection agency evaluates the weld as "yes", the output unit OUTPUT will output: Welding passed, or output via a digital display device or via printout.

[0057] If the system with neural network according to the present invention evaluates the weld as 100%, then it is no longer necessary to evaluate through an inspection agency, and the evaluation to evaluate "pass" directly enters the output unit OUTPUT, as shown in the table below.

[0058]

[0059]

[0060] Of course, it can be adapted for evaluation independently. Another consideration is that it doesn't have to reach 100% to pass; instead, the value could be 90% or another confirmed value.

[0061] Figure 2 The ultrasound scan file USSD is shown, where, as an example, both the first and second checks failed, and two checks were performed. Typically, if the first check fails, the process is interrupted after the first check of scan quality. It is clearly visible that, except for point 5, the waves on the inner and outer surfaces 1 and 2 exhibit a continuous and consistent orientation, while surface discontinuities are visible at point 5. However, the wall thickness maintains a constant orientation. Therefore, the standard and sub-standards are evaluated at specific percentages, and the first check of ultrasound scan file quality is not 100%, thus failing. However, the ultrasound scan file USSD also shows some defects in the structure, which are also outlined in boxes, where computer-implemented methods preferably identify defects by enclosed frames. Figure 2 This is the visible part. The computer-implemented method according to the invention evaluated the defects in the ultrasonic scan file USSD, where no numerical value existed, but the final value did not reach 100% in the second inspection of weld quality, and thus the ultrasonic scan file USSD was delivered to the inspection agency for visual inspection by the inspector. Even in cases with a large number of defects, even... Figure 2 The test diagram also shows that the inspector assessed the weld as failing and instructed the inspector to remove the weld and re-inspect it.

[0062] Advantageously, the quality of ultrasonic scan files is evaluated based on the standards set for the ultrasonic scans performed on the welds and pipes. Preferably, these standards include sub-standards such as the speed at which the ultrasonic scan files are created, and the identification of the zero line or the calibration of the ultrasonic scanner used to correctly detect the ultrasonic scan files. Of course, this list is not final and other sub-standards may be added.

[0063] List of reference numerals

[0064] USSD Ultrasonic Files

[0065] IN USSSD receives ultrasound scan files

[0066] 1QUSSD First Inspection Group, Ultrasound Scan Document Quality

[0067] 2QWELD Second Inspection Group, Welding Quality

[0068] WELD REP Replacement Welding

[0069] CR USSD creates ultrasound scan files

[0070] VISUAL undergoes visual evaluation by an inspection agency.

[0071] OUTPUT output evaluation

[0072] 1. Waves on the inner surface

[0073] 2. Waves on the outer surface

[0074] 3. Wall thickness

[0075] 4 defects

[0076] 5. Non-constant wall thickness orientation

Claims

1. A computer-implemented method for evaluating a non-destructive ultrasonic inspection of a plastic pipe weld, comprising the following steps: • receiving an ultrasonic scan file (USSD) by means of a server, • analyzing the ultrasonic scan file (USSD) based on predefined criteria by a computing unit, wherein the computing unit comprises a neural network, • evaluating the ultrasonic scan file (USSD) according to the predefined criteria by the computing unit.

2. The computer-implemented method of claim 1, wherein, The criteria are divided into inspection groups (1 QUSSD, 2 QWELD).

3. The computer-implemented method of claim 2, wherein, The inspection group (1 QUSSD) defines the quality of the ultrasonic scan file.

4. The computer-implemented method of claim 3, wherein, The evaluation of the quality of the ultrasonic scan file (USSD) comprises criteria regarding the wave run of the weld and the inner and / or outer surface of the pipe shown in the ultrasonic scan file (USSD).

5. The computer-implemented method of claim 3, wherein, The evaluation of the quality of the ultrasonic scan file (USSD) comprises criteria regarding the wall thickness run in the ultrasonic scan file (USSD) of the weld and the pipe.

6. The computer-implemented method of claim 3, wherein, The evaluation of the quality of the ultrasonic scan file (USSD) comprises criteria regarding the presence of a heating wire in the ultrasonic scan file (USSD).

7. The computer-implemented method of claim 2, wherein, The inspection group (2 QWELD) defines the quality of the weld.

8. The computer-implemented method of claim 1, wherein, The evaluation of the quality of the weld comprises at least one of the following criteria: anomalies, defects or inconsistent unintended structures in the ultrasonic scan file, no heating wire found.

9. The computer-implemented method according to one of claims 1 to 8, characterized in that, The ultrasonic inspection for creating the ultrasonic scan file (USSD) is performed and created by means of time-of-flight diffraction (TOFD) or phased array ultrasonic testing (PAUT).

10. A system for assessing non-destructive ultrasonic inspection of a plastic pipe weld, the system comprising: An ultrasonic scanner for performing an ultrasonic scan on a plastic pipe weld and creating an ultrasonic scan file (USSD); a server for receiving the ultrasonic scan file (USSD); a computing unit for analyzing the ultrasonic scan file (USSD) and evaluating the ultrasonic scan file based on predefined criteria, wherein the computing unit comprises a neural network; and an output unit (OUTPUT) for displaying the evaluation.

11. The system of claim 10, wherein, The server has a web-based interface.

12. A method for training a neural network for use in a method for evaluating a non-destructive ultrasonic inspection of a plastic pipe weld, preferably according to any one of claims 1 to 9, comprising the following steps: • creating simulated ultrasonic scan files (USSD) by means of the neural network of the computing unit from known ultrasonic scan files (USSD) of different materials, • evaluating the ultrasonic scan files (1 USSD, 2 QWELD) according to predefined criteria, • continuously expanding the neural network based on simulated ultrasonic scan files (USSD).

13. The method of claim 12, wherein, At least one of the following criteria can be found in the predefined criteria: the wave run shown in the ultrasonic scan file regarding the weld and the outer or inner surface of the pipe; the wall thickness run in the ultrasonic scan file of the weld and the pipe; anomalies, defects or inconsistent unintended structures in the ultrasonic scan file; no heating wire.

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

Images