A groove intelligent identification system for box-shaped steel component welding

Abstract

The application discloses a kind of bevel intelligent identification system for box type steel member welding, it is related to bevel identification technical field, it solves the technical problem that the existing steel member welding bevel identification exists blind area, and it is susceptible to noise signal influence, and identification accuracy is not high;Including: in the process of bevel processing, bevel detection module is used to carry out magnetic powder or penetration nondestructive testing to bevel using ultrasonic probe, after detection is qualified, the size information of bevel is collected by ultrasonic probe;Controller is used to compare size information with standard size information and constraint condition stored in database;If consistent, then generate size qualified signal;When receiving size qualified signal, then trigger camera module to collect image information of bevel, and the collected image information is transmitted to defect identification module for defect identification;Defect identification module is used to filter gain adjustment to preprocessed image information, to reduce signal-to-noise ratio, to improve the accuracy of bevel defect identification.

Description

Technical Field

[0001] This invention relates to the field of bevel recognition technology, specifically an intelligent bevel recognition system for welding box-type steel components. Background Technology

[0002] When processing steel components, the entire circumference of the steel pipe ends is beveled on the outer surface. During pipeline construction, the steel pipes are first assembled and then butt-welded, that is, the two steel pipes are butt-welded through the bevel. The bevels used for welding steel components generally have high requirements. During manufacturing, ultrasonic testing, magnetic particle testing, and penetrant testing are required. At the same time, due to factors such as equipment, environment, and human error, defects such as deformation, scratches, and breakage may occur in the bevel production, affecting the product performance.

[0003] Because the existing bevel wall thickness structure for welding steel components is irregular, ultrasonic probes cannot fully contact the bevel surface, resulting in blind spots. Furthermore, current computer vision detection technology is prone to image blurring when affected by noise signals, affecting the accuracy of bevel recognition. Based on these shortcomings, this invention proposes an intelligent bevel recognition system for welding box-type steel components. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an intelligent bevel recognition system for welding box-type steel components.

[0005] To achieve the above objectives, a first aspect of the present invention provides an intelligent bevel recognition system for welding box-type steel components, comprising a model training module, a bevel detection module, a controller, a database, a camera module, and a defect recognition module;

[0006] The model training module is used to collect defective product images as a sample training set, and train an LSTM neural network model to obtain a defect recognition model M; wherein, defective product images are characterized by defects such as deformation, scratches, dirt, breakage, and edge damage.

[0007] During the beveling process, the beveling inspection module is used to perform magnetic particle or penetrant non-destructive testing on the beveling using an ultrasonic probe. After the test is passed, the beveling dimension information is collected by the ultrasonic probe.

[0008] The bevel includes a first upper bevel and a second upper bevel set along the outer wall of the upper semicircle of the steel component pipe opening, and a first lower bevel and a second lower bevel set along the outer wall of the lower semicircle of the steel component pipe opening;

[0009] The bevel detection module is used to upload the collected bevel size information to the controller;

[0010] The controller is used to compare the size information with the standard size information and constraints stored in the database; if the comparison is consistent, a size qualified signal is generated; otherwise, a size unqualified signal is generated.

[0011] When a dimension pass signal is received, the camera module is triggered to acquire image information of the bevel and transmit the acquired image information to the defect identification module for identification.

[0012] The specific identification steps of the defect identification module are as follows:

[0013] The received image information is preprocessed; the filtering gain of the preprocessed image information is adjusted to reduce the signal-to-noise ratio and reduce image noise.

[0014] The adjusted image information is substituted into the defect recognition model M to identify bevel defects; when a defect is identified, a defect signal and corresponding detection data are generated; the defect recognition module is used to send the defect signal and corresponding detection data to the controller for display and storage.

[0015] Furthermore, the filtering gain of the preprocessed image information is adjusted, specifically including:

[0016] The preprocessed image information is converted into a digital signal, and the converted digital signal is filtered; the preprocessing includes sharpening, mathematical morphology transformation, binarization, edge extraction, and contour extraction.

[0017] The periodic energy value of the corresponding digital signal is collected at a preset interval and marked as KEi. The periodic energy value KEi is compared with a preset energy threshold; the preset energy threshold includes A1 and A2; where A1 < A2.

[0018] When the periodic energy value KEi ≤ A1, let the energy deviation value KLi = A1 - KEi;

[0019] When A1 < periodic energy value KEi < A2, let the energy deviation value KLi = 0;

[0020] When the periodic energy value KEi≥A2, let the energy deviation value KLi=KEi-A2;

[0021] Within a preset time period, the energy deviation value KLi is integrated over time to obtain the deviation reference index KZ; the deviation reference index KZ is then compared with a preset reference threshold.

[0022] If the deviation from the reference index KZ is greater than the preset reference threshold, an adjustment signal is generated;

[0023] When an adjustment signal is received, the gain of the digital signal is adjusted by controlling the programmable gain amplifier circuit, and the periodic energy value of the digital signal is adjusted to between the preset energy thresholds A1 and A2.

[0024] Furthermore, the specific detection steps of the bevel detection module are as follows:

[0025] During rough machining of steel components, allowance is left at the bevel position; ultrasonic probes are used to inspect from the top and side surfaces, ensuring the ultrasonic probes are in full contact with the surface of the steel components during inspection.

[0026] The steel components are beveled to the pre-welding state, and the excess material left at the bevel position is removed; magnetic particle or penetrant non-destructive testing is performed at the bevel position.

[0027] Furthermore, the dimensional information includes the angle between the first upper bevel and the outer surface of the steel component, the angle between the second upper bevel and the outer surface of the steel component, the angle between the first lower bevel and the outer surface of the steel component, the angle between the second lower bevel and the outer surface of the steel component, the length of the first upper bevel, the length of the second upper bevel, the length of the first lower bevel, and the length of the second lower bevel.

[0028] Furthermore, the constraints include: the angle between the first upper bevel and the outer surface of the steel component is less than the angle between the second upper bevel and the outer surface of the steel component; the angle between the first lower bevel and the outer surface of the steel component is less than the angle between the second lower bevel and the outer surface of the steel component; the length of the first upper bevel is greater than the length of the second upper bevel; and the length of the first lower bevel is greater than the length of the second lower bevel.

[0029] Furthermore, the periodic energy value refers to the value obtained by accumulating and averaging the energy of multiple consecutive received bits of data.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] In this invention, during the beveling process, a beveling inspection module uses an ultrasonic probe to perform magnetic particle or penetrant non-destructive testing on the beveling, reducing post-weld repairs and effectively lowering manufacturing costs. After passing the inspection, the ultrasonic probe collects the beveling's dimensional information. The controller compares this dimensional information with standard dimensional information and constraints stored in the database. If the comparison matches, a dimensional compliance signal is generated. Upon receiving the dimensional compliance signal, the camera module is triggered to collect image information of the beveling and transmits the collected image information to the defect identification module for identification. The defect identification module preprocesses the received image information and adjusts the filtering gain of the preprocessed image information to reduce the signal-to-noise ratio and image noise, thereby improving the accuracy of beveling defect identification. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a system block diagram of an intelligent bevel recognition system for welding box-type steel components according to the present invention.

[0034] Figure 2 This is a schematic diagram of the bevel used for welding box-shaped steel components in this invention. Detailed Implementation

[0035] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Please see Figures 1 to 2 The first aspect of the present invention provides an intelligent bevel recognition system for welding box-shaped steel components, including a model training module, a bevel detection module, a controller, a database, a camera module, and a defect recognition module;

[0037] The model training module is used to collect defective product images as a sample training set, and train an LSTM neural network model to obtain a defect recognition model M; among them, defective product images are characterized by defects such as deformation, scratches, dirt, breakage, and broken edges.

[0038] During the beveling process, the beveling inspection module is used to perform magnetic particle or penetrant non-destructive testing on the beveling using an ultrasonic probe. The specific steps are as follows:

[0039] During rough machining of steel components, allowance is left at the bevel position; ultrasonic probes are used to inspect from the top and side surfaces, ensuring the ultrasonic probes are in full contact with the surface of the steel components during inspection.

[0040] The steel components are beveled to the pre-welding state, and the excess material left at the bevel position is removed.

[0041] Perform magnetic particle or penetrant non-destructive testing at the bevel; avoid missed detections caused by poor ultrasonic probe contact due to structural irregularities; reduce rework after bevel welding and effectively reduce manufacturing costs;

[0042] After passing the inspection, the size information of the bevel is collected using an ultrasonic probe;

[0043] like Figure 2 As shown, the bevel for welding box-type steel components includes a first upper bevel and a second upper bevel set along the outer wall of the upper semicircle of the steel component pipe opening, and a first lower bevel and a second lower bevel set along the outer wall of the lower semicircle of the steel component pipe opening;

[0044] Therefore, the collected dimensional information includes the angle between the first upper bevel and the outer surface of the steel component, the angle between the second upper bevel and the outer surface of the steel component, the angle between the first lower bevel and the outer surface of the steel component, the angle between the second lower bevel and the outer surface of the steel component, the length of the first upper bevel, the length of the second upper bevel, the length of the first lower bevel, and the length of the second lower bevel.

[0045] The bevel detection module is used to upload the collected bevel dimension information to the controller; the controller is used to compare the dimension information with the standard dimension information and constraints stored in the database; if the comparison is consistent, a dimension qualified signal is generated; otherwise, a dimension unqualified signal is generated.

[0046] In this embodiment, the constraints include: the angle between the first upper bevel and the outer surface of the steel component is less than the angle between the second upper bevel and the outer surface of the steel component; the angle between the first lower bevel and the outer surface of the steel component is less than the angle between the second lower bevel and the outer surface of the steel component; the length of the first upper bevel is greater than the length of the second upper bevel; and the length of the first lower bevel is greater than the length of the second lower bevel.

[0047] In this embodiment, when a size qualified signal is received, the camera module is triggered to collect image information of the bevel and transmit the collected image information to the defect identification module for identification.

[0048] The specific identification steps of the defect identification module are as follows:

[0049] The received image information is preprocessed; the preprocessing includes sharpening, mathematical morphology transformation, binarization, edge extraction, and contour extraction.

[0050] The preprocessed image information undergoes filter gain adjustment to reduce the signal-to-noise ratio and image noise, thereby improving the accuracy of bevel defect identification; specifically including:

[0051] The preprocessed image information is converted into a digital signal, and the converted digital signal is then filtered.

[0052] The periodic energy value of the corresponding digital signal is collected at a preset interval and marked as KEi. The periodic energy value is the value obtained by accumulating and averaging the energy of multiple consecutive bits of data received.

[0053] The periodic energy value KEi is compared with a preset energy threshold; the preset energy threshold includes A1 and A2; where A1 < A2; when the periodic energy value KEi ≤ A1, the energy deviation value KLi = A1 - KEi is set.

[0054] When A1 < periodic energy value KEi < A2, let the energy deviation value KLi = 0;

[0055] When the periodic energy value KEi≥A2, let the energy deviation value KLi=KEi-A2;

[0056] Within a preset time period, the energy deviation value KLi is integrated over time to obtain the deviation reference index KZ; the deviation reference index KZ is then compared with a preset reference threshold.

[0057] If the deviation from the reference index KZ is greater than the preset reference threshold, an adjustment signal is generated;

[0058] When an adjustment signal is received, the gain of the digital signal is adjusted by controlling the programmable gain amplifier circuit, and the periodic energy value of the digital signal is adjusted to between the preset energy thresholds A1 and A2.

[0059] The adjusted image information is substituted into the defect recognition model M to identify bevel defects; when a defect is identified, a defect signal and corresponding detection data are generated; the defect recognition module is used to send the defect signal and corresponding detection data to the controller for display and storage.

[0060] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0061] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A smart bevel recognition system for welding box-type steel components, characterized in that, It includes a model training module, a bevel detection module, a controller, a database, a camera module, and a defect recognition module; The model training module is used to collect defective product images as a sample training set, and train an LSTM neural network model to obtain a defect recognition model M; wherein, defective product images are characterized by defects such as deformation, scratches, dirt, breakage, and edge damage. During the beveling process, the beveling inspection module is used to perform magnetic particle or penetrant non-destructive testing on the beveling using an ultrasonic probe. After the test is passed, the beveling dimension information is collected by the ultrasonic probe. The bevel includes a first upper bevel and a second upper bevel set along the outer wall of the upper semicircle of the steel component pipe opening, and a first lower bevel and a second lower bevel set along the outer wall of the lower semicircle of the steel component pipe opening; The bevel detection module is used to upload the collected bevel size information to the controller; The controller is used to compare the size information with the standard size information and constraints stored in the database; if the comparison is consistent, a size qualified signal is generated; otherwise, a size unqualified signal is generated. When a dimension pass signal is received, the camera module is triggered to acquire image information of the bevel and transmit the acquired image information to the defect identification module for identification. The specific identification steps of the defect identification module are as follows: The received image information is preprocessed; the preprocessed image information is then filtered and the gain is adjusted to reduce the signal-to-noise ratio and image noise; specifically, this includes: The preprocessed image information is converted into a digital signal, and the converted digital signal is filtered; the preprocessing includes sharpening, mathematical morphology transformation, binarization, edge extraction, and contour extraction. The periodic energy value of the corresponding digital signal is collected at a preset interval and marked as KEi. The periodic energy value KEi is compared with a preset energy threshold; the preset energy threshold includes A1 and A2; where A1 < A2. When the periodic energy value KEi ≤ A1, let the energy deviation value KLi = A1 - KEi; When A1 < periodic energy value KEi < A2, let the energy deviation value KLi = 0; When the periodic energy value KEi≥A2, let the energy deviation value KLi=KEi-A2; Within a preset time period, the energy deviation value KLi is integrated over time to obtain the deviation reference index KZ; the deviation reference index KZ is then compared with a preset reference threshold. If the deviation from the reference index KZ is greater than the preset reference threshold, an adjustment signal is generated; When an adjustment signal is received, the gain of the digital signal is adjusted by controlling the programmable gain amplifier circuit, and the periodic energy value of the digital signal is adjusted to between the preset energy thresholds A1 and A2. The adjusted image information is substituted into the defect recognition model M to identify bevel defects; when a defect is identified, a defect signal and corresponding detection data are generated; the defect recognition module is used to send the defect signal and corresponding detection data to the controller for display and storage.

2. The intelligent bevel recognition system for welding box-type steel components according to claim 1, characterized in that, The specific detection steps of the bevel detection module are as follows: During rough machining of steel components, allowance is left at the bevel position; ultrasonic probes are used to inspect from the top and side surfaces, ensuring the ultrasonic probes are in full contact with the surface of the steel components during inspection. The steel components are beveled to the pre-welding state, and the excess material left at the bevel position is removed; magnetic particle or penetrant non-destructive testing is performed at the bevel position.

3. The intelligent bevel recognition system for welding box-type steel components according to claim 1, characterized in that, The dimensional information includes the angle between the first up-bevel and the outer surface of the steel component, the angle between the second up-bevel and the outer surface of the steel component, the angle between the first down-bevel and the outer surface of the steel component, the angle between the second down-bevel and the outer surface of the steel component, the length of the first up-bevel, the length of the second up-bevel, the length of the first down-bevel, and the length of the second down-bevel.

4. The intelligent bevel recognition system for welding box-type steel components according to claim 3, characterized in that, The constraints include: the angle between the first upper bevel and the outer surface of the steel component is less than the angle between the second upper bevel and the outer surface of the steel component; the angle between the first lower bevel and the outer surface of the steel component is less than the angle between the second lower bevel and the outer surface of the steel component; the length of the first upper bevel is greater than the length of the second upper bevel; and the length of the first lower bevel is greater than the length of the second lower bevel.

5. The intelligent bevel recognition system for welding box-type steel components according to claim 1, characterized in that, The periodic energy value refers to the value obtained by accumulating and averaging the energy of multiple consecutive bits of received data.

Images