A visual inspection device for defects in a welded joint of a steam turbine rotor

Abstract

The application discloses a kind of steam turbine rotor welding place defect visual detection device, it is related to the defect detection technical field of steam turbine rotor weld, including operation platform;The present application is by constructing the defect classification rule of multi-parameter fusion, combine geometric feature with time sequence feature, can accurately distinguish different types of defects such as porosity, crack, slag inclusion, effectively avoid the misjudgment problem caused by only relying on single frame image or single feature, especially for the obvious distinguishing ability of similar morphology but different dynamic characteristics of defect;By comparing the determination result of time sequence dynamic feature with preliminary shape recognition result, form secondary determination mechanism, further improve the accuracy of defect type identification, simultaneously bind record to defect type, position, size and identification time etc., realize the traceability and data management of detection result, facilitate subsequent quality analysis and process improvement.

Description

Technical Field

[0001] This invention relates to the field of defect detection technology for turbine rotor welds, and in particular to a visual inspection device for defects in turbine rotor welds. Background Technology

[0002] As the core rotating component of a steam turbine, the quality of the welding at the turbine rotor's weld joints directly determines the rotor's structural strength and operational stability, thus affecting the safe and reliable operation of the entire steam turbine equipment. Therefore, defect detection at the turbine rotor's weld joints is a key process in the steam turbine's manufacturing, repair, and maintenance.

[0003] However, the current detection of defects in turbine rotor welds largely relies on manual labor combined with simple inspection tools. Because defects such as cracks, micro-pores, and slag inclusions in turbine rotor welds are often extremely small, and some defects are even hidden in circumferential locations within the weld, the visual resolution of the human eye has inherent limitations, making it difficult to clearly identify such minute defects. It is also impossible to accurately observe the details of welds on circumferential curved surfaces, easily leading to missed defects due to blind spots and insufficient resolution. Furthermore, significant differences in professional experience and focus among operators make it difficult to achieve completely uniform standards for judging weld defects. Some operators, due to insufficient experience, may misjudge normal weld textures as defects, or fatigue from prolonged inspections may reduce accuracy, resulting in frequent misjudgments and compromising the consistency and reliability of inspection results. Summary of the Invention

[0004] In order to solve the problems in the background art, the present invention proposes a visual inspection device for defects in the welding of steam turbine rotors.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A visual inspection device for defects in the weld of a steam turbine rotor includes an operating table, on which a rotating tooling fixture is slidably mounted via a slide rail, and an inspection device for acquiring images of the weld seam of the steam turbine rotor is mounted on the rear end face of the operating table via an installation component. A support plate is installed at one corner of the top surface of the operating table via a support member. The support plate is equipped with an image processing mechanism that works with the detection equipment to process weld images and identify defects. The image processing mechanism also includes a computer mounted on the top surface of the support plate, and the computer is equipped with a defect analysis module. The defect analysis module performs grayscale segmentation and abnormal grayscale block identification on weld images acquired by the inspection equipment. It determines the effective detection area by statistically analyzing the vertical integrity of normal grayscale blocks and stitches together multiple image frames based on rotor speed. It extracts geometric features such as area, perimeter, aspect ratio, and roundness of the defect area, and calculates temporal dynamic indicators such as area change rate, roundness change stability, aspect ratio change stability, and grayscale change stability. Combining geometric and temporal features, it classifies defects into porosity, cracks, or inclusions according to preset threshold rules. After comparing the judgment results with the preliminary identification results, it binds and records information such as defect type, location, size, and identification time, achieving automatic weld defect identification and data management.

[0006] Preferably, the effective area analysis steps of the defect analysis module are as follows: M1: Performs grayscale processing on the real-time acquired image data and divides the grayscale image into segments according to a preset pixel block size. Each grayscale block is of the same size and is numbered according to its row and column number in the image; within each grayscale block of the image acquired at the same time... Calculate the mean of each gray value and standard deviation Set the grayscale fluctuation range to Gray values ​​outside this range are marked as outliers, and the number of outliers is recorded. ; M2: For those that satisfy The grayscale blocks are marked as normal grayscale blocks and their numbers are retained; after all grayscale blocks have been identified, the number of normal grayscale blocks in each vertical direction is counted. ,like If the image data in that column is complete, then the column number is recorded. The total number of columns for grayscale blocks in the segmented image data; for those satisfying The column numbers are retrieved. If the column numbers are consecutive, the consecutive counts are recorded. The effective area at the corresponding time is determined by combining the average consecutive counts of the image data at each moment during the rotor rotation process in the historical data.

[0007] Preferably, the defect analysis steps of the defect analysis module are as follows: N1: During rotor rotation, the rotor speed is recorded. The number of columns used for image stitching is determined according to the number of columns corresponding to the effective area, and the overall image data of the rotor surface is stitched together. Abnormal grayscale blocks are judged on the stitched image. Abnormal grayscale blocks are colored and their shapes are depicted. Based on the length and width dimensions of the shape, the defects are initially identified. Irregular shapes are initially judged as slag inclusions, round shapes are initially judged as pores, and aspect ratios exceeding the preset threshold are initially judged as cracks. N2: Acquisition Frame image, in the Calculate the circularity of the defect area in the frame image. and aspect ratio Calculate the rate of change of area Stability of circularity variation Stability of aspect ratio variation and grayscale change stability Defect type is determined based on preset threshold rules: N3: If and and and If so, the defect type of the corresponding defect area is determined to be porosity; if and and If so, the defect type in the corresponding defect area is determined to be a crack; if or If the defect type in the corresponding defect area is determined to be slag inclusion, then... and These are the average values ​​for roundness and aspect ratio, respectively. The circularity threshold. The stomatal aspect ratio threshold. and These are the aspect ratio and roundness threshold of the crack, respectively. and These are the grayscale change thresholds for pores and inclusions, respectively. , and These are the stability thresholds for roundness, crack aspect ratio, and inclusion aspect ratio, respectively. The judgment result is compared with the preliminary identification result. If the two judgment results are consistent, the defect type, location, size, and identification time are bound and recorded.

[0008] Preferably, the image processing mechanism further includes a keyboard and mouse adapted to the computer, the detection device is electrically connected to the computer via wires, and the computer has a built-in processing program for analyzing weld images and identifying defects.

[0009] Preferably, the support includes a base fixed to the top surface of the operating table, a sleeve fixed to the top surface of the base, and a movable rod slidably disposed in the sleeve. The top of the movable rod is fixed to the bottom surface of the support plate. The top of the sleeve is provided with a threaded hole, and a locking bolt is threaded into the threaded hole.

[0010] Preferably, the mounting component includes an electric actuator fixedly mounted on the rear end of the top surface of the operating table, with a bracket fixedly mounted on the movable end of the electric actuator, and the detection equipment fixedly mounted on the bracket.

[0011] Preferably, a rubber pad is fixedly installed at the front end of the top surface of the support plate.

[0012] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. The vertical height of the testing equipment is adjusted by the extension and retraction of the electric actuator, enabling the equipment to be precisely aligned with the welding points of turbine rotors of different specifications, ensuring the accuracy of the target surface for image acquisition. At the same time, the support components of the image processing mechanism adopt an adjustable structure of sleeve and movable rod, which can adjust the height of the support plate according to the operator's operating habits and visual comfort. It can also adapt to the installation height requirements of different models of image processing equipment, greatly improving the ergonomics of the testing operation and reducing operator fatigue. 2. By constructing a multi-parameter fusion defect classification rule through the defect analysis module, combining geometric features with temporal features, it can accurately distinguish different types of defects such as porosity, cracks, and inclusions, effectively avoiding misjudgment problems caused by relying solely on a single frame image or a single feature. It has a particularly strong ability to distinguish defects with similar shapes but different dynamic characteristics. By comparing the judgment results of temporal dynamic features with the preliminary shape recognition results, a secondary judgment mechanism is formed to further improve the accuracy of defect type identification. At the same time, information such as defect type, location, size, and recognition time are bound and recorded to achieve traceability and data management of the detection results, which facilitates subsequent quality analysis and process improvement. Attached Figure Description

[0013] Figure 1 A schematic diagram of the overall structure provided according to an embodiment of the present invention is shown; Figure 2 A structural schematic diagram of a support member provided according to an embodiment of the present invention is shown; Figure 3 An exploded view of a support member provided according to an embodiment of the present invention is shown; Figure 4 A system flowchart according to an embodiment of the present invention is shown.

[0014] Legend: 1. Operating table; 2. Rotary tooling fixture; 3. Electric actuator; 4. Testing equipment; 5. Slide rail; 6. Support component; 601. Base; 602. Sleeve; 603. Movable rod; 604. Locking bolt; 605. Threaded hole; 7. Computer; 8. Keyboard; 9. Mouse; 10. Bracket; 11. Support plate; 12. Rubber pad. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0016] Example 1: Please refer to Figure 1 - Figure 3 The present invention provides a technical solution: A visual inspection device for weld defects in a steam turbine rotor includes an operating table 1. A rotating fixture 2 (existing technology) is slidably mounted on the operating table 1 via a slide rail 5. Specifically, the slide rail 5 is symmetrically fixed to the rotating fixture 2, and the rotating fixture 2 is slidably positioned at both ends of the slide rail 5. A drive unit (existing technology, not shown in the figure) is provided within the operating table 1 to drive the rotating fixture 2 to move relative to each other, thereby facilitating the clamping and fixing of the steam turbine rotor. The rotating fixture 2 can drive the clamped steam turbine rotor to rotate around its own axis, allowing the inspection equipment to capture images of the circumferential surface of the weld joint of the steam turbine rotor, achieving full circumferential inspection of the weld without blind spots and avoiding blind spots caused by rotor fixation. An inspection device 4 for acquiring images of the steam turbine rotor weld joint is mounted on the rear end face of the operating table 1 via a mounting component. The inspection device 4 is an industrial high-definition camera used to acquire images of the steam turbine rotor weld joint.

[0017] A support plate 11 is installed at one corner of the top surface of the operating table 1 via a support member 6. The support member 6 provides adjustable support for the support plate 11. The support plate 11 is equipped with an image processing mechanism that works with the inspection equipment 4 to process weld images and identify defects.

[0018] In this invention, the image processing mechanism includes a computer 7 mounted on the top surface of the support plate 11, a keyboard 8 and a mouse 9 adapted to the computer 7. The keyboard 8 and mouse 9 serve as human-computer interaction components, facilitating the operator to operate the computer's detection program, set parameters, and view, edit, and export the detection results. The detection device 4 is electrically connected to the computer 7 via wires. The turbine rotor weld images acquired by the detection device 4 can be transmitted to the computer 7 in real time via wires. The computer 7 has a built-in processing program specifically for analyzing weld images and identifying defects. This program can perform image algorithm processing on the acquired high-definition weld images, automatically identify defects such as cracks, porosity, slag inclusions, and lack of fusion in the weld, and mark and statistically analyze the location, size, and type of defects. Compared with the traditional method of manually identifying defects, this significantly improves the accuracy and efficiency of defect identification, while also enabling digital storage of detection data, facilitating subsequent traceability and analysis of detection records.

[0019] In this invention, the support member 6 includes a base 601 fixed to the top surface of the operating table 1, which provides a stable mounting foundation for the support member and prevents it from tipping over; a sleeve 602 fixed to the top surface of the base 601 and a movable rod 603 slidably disposed within the sleeve 602; the top of the movable rod 603 is fixed to the bottom surface of the support plate 11; a threaded hole 605 is provided on the top of the sleeve 602, and a locking bolt 604 is threaded into the threaded hole 605; when the movable rod 603 is adjusted to a specified height, the locking bolt 604 is tightened, and its end abuts against the outer wall of the movable rod 603, thereby fixing the movable rod 603 relative to the sleeve 602 and locking the support plate 11 at the specified height. This adjustable structure of the support member allows the height of the support plate 11 to be adjusted according to the operator's operating habits and visual comfort, improving the ergonomics of the inspection operation, and also adapting to the installation height requirements of different models of image processing equipment.

[0020] In this invention, the mounting component includes an electric push rod 3 fixedly installed at the rear end of the top surface of the operating table 1. A bracket 10 is fixedly installed at the movable end of the electric push rod 3, and the detection device 4 is fixedly installed on the bracket 10. The extension and retraction of the electric push rod 3 can drive the bracket 10 and the detection device 4 to complete the vertical height adjustment, which enables the detection device 4 to accurately align with the welding point of the turbine rotor and ensure the accuracy of the target surface for image acquisition.

[0021] In this invention, a rubber pad 12 is fixedly installed on the front end of the top of the support plate 11. The rubber pad 12 is made of elastic and wear-resistant rubber material. By using the rubber pad 12, the friction between the keyboard 8 and the support plate 11 can be increased, and the keyboard 8 can be prevented from sliding off the support plate 11, thereby facilitating the protection of the keyboard 8.

[0022] Working principle: When using this invention, the turbine rotor to be tested is first placed between the rotating fixtures 2, and then the rotating fixtures 2 are fixed to the turbine rotor by the power drive unit. The vertical height of the testing equipment 4 is adjusted by the electric actuator 3 so that the testing equipment 4 is accurately aligned with the weld of the turbine rotor; at the same time, according to the operator's operating habits, the movable rod 603 of the support component 6 is adjusted to adjust the support plate 11 to a suitable height and the locking bolt 604 is tightened to lock it. Start the testing equipment 4 and the rotating fixture 2. The rotating fixture 2 drives the turbine rotor to rotate slowly. The testing equipment 4 simultaneously performs circumferential high-definition image acquisition on the rotor welding joint and transmits the acquired images to the computer 7 in real time. The built-in processing program of computer 7 automatically analyzes and identifies defects in the transmitted weld images, marks the identified defects and performs data statistics. Operators can view the inspection results on the computer using keyboard 8 and mouse 9, and record and export the defect data. Inspection completion: After the inspection is completed, turn off the power to all equipment, loosen the rotating fixture 2, remove the inspected turbine rotor, clean the operating table and all components, and complete the entire inspection process.

[0023] Example 2: Please refer to Figure 4 The computer 7 has a defect analysis module built in. The defect analysis module performs grayscale segmentation and abnormal grayscale block identification on weld images acquired by the inspection equipment. It determines the effective detection area by statistically analyzing the vertical integrity of normal grayscale blocks and stitches together multiple image frames based on rotor speed. It extracts geometric features such as area, perimeter, aspect ratio, and roundness of the defect area, and calculates temporal dynamic indicators such as area change rate, roundness change stability, aspect ratio change stability, and grayscale change stability. Combining geometric and temporal features, it classifies defects into porosity, cracks, or inclusions according to preset threshold rules. After comparing the judgment results with the preliminary identification results, it binds and records information such as defect type, location, size, and identification time, achieving automatic weld defect identification and data management. The real-time acquired image data is processed into grayscale, and the grayscale image is segmented according to the size of pixel blocks, resulting in... The image data consists of several identical grayscale blocks, numbered according to their row and column numbers on the grayscale image. The acquired image data is sorted by acquisition time, and the corresponding numbered grayscale blocks within a single image acquired at the same time are further analyzed. Average the gray values and standard deviation The calculation, and the mean obtained from the calculation. and standard deviation Range of grayscale data The setting compares the corresponding grayscale value data with the corresponding grayscale value fluctuation range, marks the corresponding grayscale value data that is outside the fluctuation range as an outlier, and records the number of outliers. ; To satisfy Grayscale blocks are marked as normal grayscale blocks, and their numbers are recorded. After determining all numbered grayscale blocks in the entire image data, only the normal grayscale blocks with corresponding numbers are retained in the image data. The number of normal grayscale blocks retained in each vertical direction is recorded. If statistics are performed, If so, it is determined that the column remains intact in the image data, and the column number is recorded. The total number of columns for grayscale blocks in the segmented image data; for those satisfying The column numbers are retrieved. If the column numbers are consecutive, the consecutive counts are recorded. Then, the corresponding consecutive counts are obtained from the image data acquired at each moment during the rotor rotation process in the historical data. The average of the consecutive counts is taken, and the continuous area corresponding to the average of the consecutive counts at the corresponding moment is determined as the valid area. During the rotor's rotation, the rotor speed is recorded, and the column number corresponding to the speed at that moment is calculated. This column number is compared with the column number corresponding to the effective area. If the column number corresponding to the speed is greater than the column number corresponding to the effective area, the column number corresponding to the effective area is used. If the column number corresponding to the speed is less than the column number corresponding to the effective area, the column number corresponding to the speed is used. Based on the obtained column number, the overall image data of the rotor surface is stitched together. After stitching, abnormal grayscale blocks are identified on the stitched complete image. Then, the abnormal grayscale blocks are colored, and the shape of the colored area is depicted. Based on the length and width dimensions of the shape, defects such as cracks, pores, and inclusions are identified, and the location data of the corresponding defects on the rotor surface image (the number of the abnormal grayscale block on the overall image) is recorded. Irregular shapes are initially identified as inclusions, round shapes are initially identified as pores, and aspect ratios exceeding a preset threshold are initially identified as cracks. Get Frame image, in the In the frame image, the circularity of the defect area , For the perimeter, The area of ​​the defective region; the aspect ratio of the defective region. , and These represent the long and short sides of the circumscribed rectangle of the defect region, respectively; and the rate of change of area during rotation. , In the first Area of ​​defective regions in frame images; stability of circularity variation mean roundness Stability of aspect ratio variation Average aspect ratio The average gray level of the defect area is grayscale change stability , In the first The average gray level of the defective region in the frame image; like and and and If so, the defect type of the corresponding defect area is determined to be porosity; if and and If so, the defect type in the corresponding defect area is determined to be a crack; if or If the defect type in the corresponding defect area is determined to be slag inclusion, then... The circularity threshold. The stomatal aspect ratio threshold. and These are the aspect ratio and roundness threshold of the crack, respectively. and These are the grayscale change thresholds for pores and inclusions, respectively. , and These are the stability thresholds for roundness, crack aspect ratio, and inclusion aspect ratio, respectively. After initially judging the defects on the rotor surface through grayscale, the defects are then judged again by the temporal dynamic characteristics during rotor rotation. If the two judgments are consistent, the defect type, location, size and other data are bound and recorded with the corresponding identification time for easy retrieval by staff.

[0024] The above description of the embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A visual inspection device for defects in the welded joints of a steam turbine rotor, comprising an operating table (1), characterized in that, A rotating tooling fixture (2) is slidably provided on the operating table (1) via a slide rail (5), and a detection device (4) for collecting images of the turbine rotor weld is installed on the rear end face of the operating table (1) via an installation component. A support plate (11) is installed at one corner of the top surface of the operating table (1) via a support member (6). The support plate (11) is equipped with an image processing mechanism that works in conjunction with the detection equipment (4) to process weld images and identify defects. The image processing mechanism also includes a computer (7) mounted on the top surface of the support plate (11), and the computer (7) is equipped with a defect analysis module. The defect analysis module performs grayscale segmentation and abnormal grayscale block identification on weld images acquired by the inspection equipment. It determines the effective detection area by statistically analyzing the vertical integrity of normal grayscale blocks and stitches together multiple image frames based on rotor speed. It extracts geometric features such as area, perimeter, aspect ratio, and roundness of the defect area, and calculates temporal dynamic indicators such as area change rate, roundness change stability, aspect ratio change stability, and grayscale change stability. Combining geometric and temporal features, it classifies defects into porosity, cracks, or inclusions according to preset threshold rules. After comparing the judgment results with the preliminary identification results, it binds and records information such as defect type, location, size, and identification time, achieving automatic weld defect identification and data management.

2. The visual inspection device for weld defects at turbine rotors according to claim 1, characterized in that, The effective area analysis steps of the defect analysis module are as follows: M1: Performs grayscale processing on the real-time acquired image data and divides the grayscale image into segments according to a preset pixel block size. Each grayscale block is of the same size and is numbered according to its row and column number in the image; within each grayscale block of the image acquired at the same time... Calculate the mean of each gray value and standard deviation Set the grayscale fluctuation range to Gray values ​​outside this range are marked as outliers, and the number of outliers is recorded. ; M2: For those that satisfy The grayscale blocks are marked as normal grayscale blocks and their numbers are retained; After determining all grayscale blocks, count the number of normal grayscale blocks in each vertical direction. ,like If the image data in that column is complete, then the column number is recorded. The total number of columns for grayscale blocks in the segmented image data; for those satisfying The column numbers are retrieved. If the column numbers are consecutive, the consecutive counts are recorded. The effective area at the corresponding time is determined by combining the average consecutive counts of the image data at each moment during the rotor rotation process in the historical data.

3. The visual inspection device for weld defects at turbine rotors according to claim 1, characterized in that, The defect analysis steps in the defect analysis module are as follows: N1: During the rotor rotation process, record the rotor speed, determine the number of columns used for image stitching based on the number of columns corresponding to the effective area, and complete the stitching of the overall image data of the rotor surface. Abnormal grayscale blocks are identified on the stitched complete image. The abnormal grayscale blocks are colored and their shapes are depicted. Based on the length and width of the shape, the defects are initially identified. Irregular shapes are initially identified as slag inclusions, round shapes are initially identified as pores, and aspect ratios exceeding a preset threshold are initially identified as cracks. N2: Acquisition Frame image, in the Calculate the circularity of the defect area in the frame image. and aspect ratio Calculate the rate of change of area Stability of circularity variation Stability of aspect ratio variation and grayscale change stability Defect type is determined based on preset threshold rules: N3: If and and and If so, the defect type of the corresponding defect area is determined to be porosity; if and and If so, the defect type in the corresponding defect area is determined to be a crack; if or If the defect type in the corresponding defect area is determined to be slag inclusion, then... and These are the average values ​​for roundness and aspect ratio, respectively. The circularity threshold. The stomatal aspect ratio threshold. and These are the aspect ratio and roundness threshold of the crack, respectively. and These are the grayscale change thresholds for pores and inclusions, respectively. , and These are the stability thresholds for roundness, crack aspect ratio, and inclusion aspect ratio, respectively. The judgment result is compared with the preliminary identification result. If the two judgment results are consistent, the defect type, location, size, and identification time are bound and recorded.

4. The visual inspection device for weld defects at turbine rotors according to claim 1, characterized in that, The image processing mechanism also includes a keyboard (8) and a mouse (9) adapted to the computer (7). The detection device (4) is electrically connected to the computer (7) via wires. The computer (7) has a built-in processing program for analyzing weld images and identifying defects.

5. The visual inspection device for weld defects at turbine rotors according to claim 1, characterized in that, The support member (6) includes a base (601) fixed on the top surface of the operating table (1), a sleeve (602) fixed on the top surface of the base (601), and a movable rod (603) slidably disposed in the sleeve (602). The top of the movable rod (603) is fixedly connected to the bottom surface of the support plate (11). The top of the sleeve (602) is provided with a threaded hole (605), and a locking bolt (604) is threaded into the threaded hole (605).

6. The visual inspection device for weld defects at turbine rotors according to claim 1, characterized in that, The mounting component includes an electric push rod (3) fixedly installed at the rear end of the top surface of the operating table (1), and a bracket (10) is fixedly installed at the movable end of the electric push rod (3). The detection device (4) is fixedly installed on the bracket (10).

7. The visual inspection device for weld defects at turbine rotors according to claim 1, characterized in that, A rubber pad (12) is fixedly installed on the front end of the top of the support plate (11).

Images