Ultrasonic combined visual weld defect detection robot and detection method thereof

Abstract

The application discloses a kind of ultrasonic combination visual weld defect detection robot and its detection method, robot includes the mobile platform of Mecanum wheel and weld defect detection unit;Mecanum wheel mobile platform includes tail frame, middle frame and Mecanum wheel device;Mecanum wheel device includes motor frame, magnetic attraction unit, Mecanum wheel drive motor and Mecanum wheel;Tail frame is installed in the way of adjustable pitch angle;Motor frame of Mecanum wheel device is installed in the way of adjustable position;Weld defect detection unit includes ultrasonic detection module and visual detection module.Detection method includes simultaneously using ultrasonic detection module to detect, and signal is handled;Using visual detection module to detect, and extracting defect surface image texture feature;The features after processing are fused, and the three-dimensional topography of weld defect is reconstructed and obtained;The detection robot of the application can realize the deep fusion of ultrasonic and visual technology, combined with flexible mobile platform, the appearance and internal defect of storage tank weld are efficiently, accurately and comprehensively detected.

Description

Technical Field

[0001] This invention relates to the field of weld defect inspection equipment, and more specifically to a weld defect inspection robot using a combination of ultrasound and vision. Background Technology

[0002] Weld quality inspection is a core component of safety assessments for industrial equipment such as pressure vessels and oil storage tanks. Current mainstream inspection technologies mainly fall into three categories: ultrasonic testing, radiographic testing, and visual inspection, but all have significant technical bottlenecks. Ultrasonic testing relies on high-frequency sound waves to analyze echo signals and identify internal defects such as porosity and inclusions. However, traditional single-probe testing suffers from slow scanning speeds, high dependence on operator experience, and low coverage of complex geometric welds. Radiographic testing uses X-rays to penetrate the workpiece and image it, directly displaying defects such as cracks and lack of fusion. However, existing equipment generally weighs over 80 kg, requiring a dedicated protective enclosure, making on-site mobile inspection difficult. Machine vision inspection technology acquires weld appearance images based on image sensors and identifies surface defects through edge detection algorithms. However, in actual industrial environments, factors such as metal surface reflection and oil contamination lead to a high false alarm rate.

[0003] Existing weld inspection technologies are typically based on a single method, making it difficult to simultaneously meet the comprehensive inspection needs for both external and internal defects. Ultrasonic and radiographic techniques focus only on internal defects, making it difficult to accurately locate the positional relationship between external and internal defects; visual inspection lacks the ability to analyze the correlation between internal defects, leading to difficulties in diagnosing the causes of defects; and the separate use of multiple devices creates data silos.

[0004] On the other hand, traditional defect detection robots equipped with tracked / wheeled chassis lack sufficient surface fit. Most existing inspection robots use tracked or ordinary wheeled chassis, relying on the contact friction between the tracks or wheels and the wall for drive. Tracked chassis must maintain close contact with the wall at all times, but tank welds often have abrupt curvature changes (such as corners, arc transition areas) or continuous curvature variations. The rigid structure of the tracks cannot adapt to surface deformation in real time, easily leading to slippage, jamming, or suspension, causing deviation from the inspection path. Although magnetic wheel wall-climbing robots are fixed by magnetic adsorption, the spacing between the magnetic wheels is fixed. When the curvature of the tank wall changes, the normal angle between the magnetic wheels and the wall deviates, the adsorption force weakens, and there is a risk of tipping over. Furthermore, traditional ultrasonic inspection robots mostly use single-point probes or fixed arrays. When the curvature of the tank wall changes, the normal angle between the probe and the weld cannot be dynamically adjusted, resulting in deviation of the ultrasonic incident angle, attenuation of the echo signal, and an increased false negative rate. Summary of the Invention

[0005] In view of this, the purpose of this invention is to develop an ultrasonic combined with vision weld defect detection robot, which realizes the deep integration of ultrasonic and vision technologies, and combines a flexible mobile platform to perform efficient and accurate comprehensive detection of the appearance and internal defects of tank welds.

[0006] The ultrasonic combined vision weld defect detection robot of the present invention includes a highly adaptable McLaren mobile platform and a weld defect detection unit disposed at the bottom of the highly adaptable McLaren mobile platform;

[0007] The highly adaptable Mecanum wheel mobile platform includes a rear frame, a middle frame, and a Mecanum wheel assembly. The Mecanum wheel assembly includes a motor frame, a magnetic unit located at the bottom of the motor frame, a Mecanum wheel drive motor fixedly installed inside the motor frame, and Mecanum wheels driven to rotate by the Mecanum wheel drive motor. The rear frame is connected to the front and rear sides of the middle frame in a pitch-adjustable manner. The motor frame of the Mecanum wheel assembly is connected to the left and right sides of the rear frame in a position-adjustable manner.

[0008] The weld defect detection unit includes an ultrasonic inspection module and a visual inspection module installed on the central frame.

[0009] Preferably, the motor frame of the wheat wheel device is connected to the rear frame via a quadrilateral linkage mechanism, and the rear frame is provided with a wheat wheel adjustment mechanism for driving the quadrilateral linkage mechanism to deform and change the position of the wheat wheel device.

[0010] Preferably, the rear frame and the middle frame are connected by a number of upper and lower connecting rods. One end of the lower connecting rod is hinged to the middle frame, and the other end is slidably fitted into the strip groove of the middle frame. Both ends of the upper connecting rod are slidably fitted into the strip grooves of the rear frame and the middle frame, respectively. A spring is provided in the strip groove to spring against the sliding end of the connecting rod.

[0011] Preferably, the rear frame includes a base plate and adjustment frames hinged to both sides of the base plate; a plurality of connecting rods are hinged between the motor frame and the adjustment frames, so that the adjustment frames, connecting rods and motor frame together form a quadrilateral linkage mechanism.

[0012] Preferably, the wheel adjustment mechanism includes a screw and nut pair disposed on the rear frame, a slider driven to move by the nut, and a quadrilateral connecting frame; one side of the quadrilateral connecting frame is hinged to the slider, and the other side is hinged to the adjustment frame and fixedly connected to the connecting rod; the slider, driven by the screw and nut pair, drives the connecting rod to swing through the quadrilateral connecting frame, so that the quadrilateral connecting rod mechanism deforms to complete the position adjustment of the wheel device.

[0013] Preferably, the ultrasonic detection module includes a probe shaft, an ultrasonic probe, a probe housing, and a probe outer wheel; the probe shaft is horizontally mounted at the bottom of the central frame, and several ultrasonic probes are distributed on the probe shaft to form an ultrasonic probe array; the probe housing is mounted on the probe shaft via bearings and surrounds the ultrasonic probes, and the two sides of the probe housing are sealed with sealing rings and filled with coupling fluid; the probe outer wheel is fixedly fitted onto the probe housing.

[0014] Preferably, the motor frame is supported by a top plate via a shock absorber, with the lower end of the shock absorber hinged to the motor frame and the upper end hinged to a shock absorber mounting base at the bottom of the top plate.

[0015] Preferably, a knob frame is mounted on the bottom of the top plate via a bracket, and the knob frame is equipped with an adjustment knob fixed to the upper end of the lead screw.

[0016] Preferably, the visual inspection module includes a high-resolution camera disposed on the central frame.

[0017] This invention also discloses a method for detecting weld defects using the aforementioned inspection robot, characterized by comprising the following steps:

[0018] S1. The robot starts and positions the weld seam using a highly adaptable mobile platform;

[0019] S2. Capture the probe's motion trajectory, simultaneously perform detection using the ultrasonic testing module, and process the ultrasonic time-domain and frequency-domain signals; perform detection using the visual testing module, and extract the texture features of the defect surface image;

[0020] S3. The processed ultrasonic signal features are fused with the extracted visual texture features, and the three-dimensional morphology of the weld defect is obtained by point cloud reconstruction.

[0021] S4. Defect detection and classification of the three-dimensional morphology are performed using deep learning algorithms;

[0022] S5. Generate a diagnostic report.

[0023] The beneficial effects of this invention are:

[0024] 1. The ultrasonic testing module of this invention has ultrasonic probes arranged in an array along the probe axis, covering the width and depth of the weld; and adopts dry coupling testing technology to avoid the limitations of coupling fluid on on-site operation.

[0025] 2. The visual inspection module in this invention uses a high-resolution camera to perform a comprehensive scan of the weld appearance. Combined with existing deep learning algorithms, it can automatically identify surface defects such as undercut, weld beads, and incomplete welding.

[0026] 3. When the ultrasonic testing module and the visual testing module work synchronously in this invention, a full-dimensional scan of the weld surface and interior can be completed in one testing cycle, avoiding the omission of complementary defects by traditional single technology. The location of the external defect is compared with the internal defect data to output a comprehensive test result.

[0027] 4. The highly adaptable wheel-mounted moving platform used in this invention allows the wheel device to move freely 360° in a plane via obliquely distributed rollers, moving along any trajectory such as the circumferential weld seam or longitudinal weld seam of the storage tank without steering. Furthermore, in this invention's wheel-mounted moving platform, the position of the wheel relative to the rear frame can be adjusted, and the pitch angle of the rear frame relative to the middle frame can be adjusted. Therefore, the contact angle between the wheel and the curved surface can be adjusted according to the curvature of the storage tank wall, increasing the contact area. Simultaneously, by changing the position of the wheel device relative to the frame and the pitch angle of the rear frame relative to the middle frame, the distance and angle between the probe and the surface being inspected can be adjusted, ensuring effective detection. Attached Figure Description

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0029] Figure 1 This is a schematic diagram of the overall structure of the present invention. Figure I ;

[0030] Figure 2 This is a schematic diagram of the overall structure of the present invention. Figure II ;

[0031] Figure 3 This is a schematic diagram showing the connection between the middle frame and the rear frame of the present invention;

[0032] Figure 4 This is a schematic diagram of the ultrasonic detection module of the present invention;

[0033] Figure 5 This is a cross-sectional view of the ultrasonic detection module of the present invention;

[0034] Figure 6 This is a flowchart of the weld defect detection method of the present invention.

[0035] Reference numerals: Top plate-1, Bracket-2, Shock absorber-3, Motor frame-4, Mecanum wheel-5, Mecanum wheel drive motor-6, Connecting rod-7, Adjustment frame-8, Base plate-9, Lead screw-10, Quadrilateral connecting frame-11, Ultrasonic detection module-12, Visual inspection module-13, Slider-14, Knob bracket-15, Adjustment knob-16, Upper connecting rod-17, Lower connecting rod-18, Strip groove-19, Middle frame-20, Rear frame-21, Ultrasonic probe-121, Detector outer wheel-122, Bearing-123, Sealing ring-124, Detector housing-125. Detailed Implementation

[0036] Example 1

[0037] like Figure 1 , 2 As shown in Figures 3 and 4, the ultrasonic combined vision weld defect detection robot of this embodiment includes a highly adaptable McLaren mobile platform and a weld defect detection unit disposed at the bottom of the highly adaptable McLaren mobile platform.

[0038] The highly adaptable Mecanum wheel mobile platform includes two rear frames 21, a central frame 20 located between the two rear frames 21, and Mecanum wheel devices located on both sides of the central frame 20. Each Mecanum wheel device includes a motor frame 4, a magnetic attraction unit located at the bottom of the motor frame 4, a Mecanum wheel drive motor 6 fixedly installed within the motor frame 4, and Mecanum wheels 5 driven by the Mecanum wheel drive motor 6. The Mecanum wheel 5 mainly consists of a hub and rollers. The hub is the core supporting component of the wheel, and the rollers are evenly distributed along the edge of the hub, with their axes forming a certain angle with the hub axis. The Mecanum wheel 5 can achieve omnidirectional movement, including forward, backward, lateral, diagonal, and stationary rotation. This allows it to flexibly maneuver and accurately position itself within narrow pipes, greatly improving operational efficiency and flexibility. The magnetic attraction unit uses existing permanent magnets or electromagnets to reliably attach the Mecanum wheel to the surface being inspected.

[0039] The motor frame 4 is used to fix and mount the wheel drive motor 6. It has a motor positioning mounting seat inside. The wheel drive motor 6 is a servo motor, and its shaft is connected to the Mecanum wheel 5 via a coupling. The rear frame 21 includes a base plate 9 and adjustment frames 8 hinged to both sides of the base plate 9. Two pairs of connecting rods 7 are hinged between the motor frame 4 and the adjustment frames 8, forming a quadrilateral linkage mechanism with the adjustment frames 8, connecting rods 7, and motor frame 4. The rear frame 21 is equipped with a wheel adjustment mechanism for driving the quadrilateral linkage mechanism to deform and change the position of the wheel device. The wheel adjustment mechanism includes a screw and nut pair installed on the rear frame 21, a slider 14 driven by the nut to move horizontally, and a quadrilateral connecting frame 11. One side of the quadrilateral connecting frame 11 is hinged to the slider 14, and the other side is hinged to the upper side of the adjustment frame 8 and fixed to one end of the upper connecting rod 7 in the quadrilateral linkage mechanism. Driven by the screw and nut pair, the slider 14 drives the upper connecting rod 7 in the quadrilateral linkage mechanism to swing through the quadrilateral connecting frame 11, so that the quadrilateral linkage mechanism deforms to complete the position adjustment of the wheel device, and finally realizes the adjustment of the chassis height and the height of the weld defect detection unit.

[0040] The motor frame 4 is supported by a top plate 1 via shock absorbers 3. The lower end of the shock absorber 3 is hinged to the motor frame 4, and the upper end is hinged to a shock absorber 3 fixing seat at the bottom of the top plate 1. The top plate 1 is supported by four shock absorbers 3, which reduce the transmission of vibration to the top plate 1, keeping the top plate 1 relatively stable. Other detection or positioning equipment can be installed on the top plate 1 as needed. A knob bracket 15 is installed at the bottom of the top plate 1 via a bracket 2. The knob bracket 15 has an adjustment knob 16 fixed to the upper end of the lead screw 10. Before conducting the detection operation, the user can rotate the adjustment knob 16 to rotate the lead screw 10 according to the shape and size of the pipe, thereby adjusting the relative height of the wheel and the frame.

[0041] like Figure 3 The rear frame 21 and the middle frame 20 are connected by two sets of upper connecting rods 17 and lower connecting rods 18. One end of the lower connecting rod 18 is hinged to the middle frame 20, and the other end is slidably engaged with a strip groove 19 on the middle frame 20 via a slider. Both ends of the upper connecting rod 17 are equipped with sliders that slidably engage with the strip grooves 19 on the rear frame 21 and the middle frame 20, respectively. A spring is installed in the strip groove 19, and its elastic force pushes the sliders at the sliding ends of the upper connecting rod 17 and the lower connecting rod 18 to the end position of the strip groove 19. When the robot encounters unevenness or curvature changes on the surface of the tank during movement, the spring in the strip groove 19 will be compressed under external force, allowing the rear frame 21 to adaptively adjust its pitch relative to the middle frame 20. This enables the robot to better conform to the weld seam during movement, ensuring the stability of the distance and angle between the detection probe and the weld seam surface. For example, when encountering localized bulges or depressions in the weld, the chassis can automatically adjust its pitch angle to reduce the angular change of the ultrasonic array probe relative to the surface being inspected during robot movement. This maintains optimal inspection results and avoids data deviation caused by probe tilt.

[0042] like Figure 4 , 5As shown, in this embodiment, the ultrasonic testing module 12 includes a probe shaft, an ultrasonic probe 121, a probe housing 125, and a probe outer wheel 122. The probe shaft is horizontally mounted on the bottom of the central frame 20, and two pairs of ultrasonic probes 121 are mounted on the probe shaft perpendicular to its axis. The probe housing 125 is mounted on the probe shaft via bearings 123 and surrounds the ultrasonic probe 121. The two sides of the probe housing 125 are sealed with sealing rings 124 and then filled with coupling fluid. The probe outer wheel 122 is fixedly fitted onto the probe housing 125. In traditional ultrasonic testing, coupling fluid needs to fill the tiny gaps between the ultrasonic probe and the surface of the object being tested to reduce the obstruction of ultrasonic wave propagation by air gaps, allowing ultrasonic waves to smoothly penetrate into the interior of the material being tested. However, the use of coupling fluid presents many inconveniences. At large storage tank weld inspection sites, the environment is complex, and the use of coupling fluid can easily cause pollution, soiling equipment and the work area, increasing cleaning costs and time. Applying and cleaning the coupling fluid is cumbersome; it needs to be applied evenly before inspection and thoroughly cleaned afterward, which not only consumes a lot of manpower but also extends the inspection cycle. In this embodiment, the coupling fluid is encapsulated within the probe housing 125: during inspection, the probe wheel contacts the surface of the workpiece, eliminating the need to apply coupling agent to the surface. This avoids the pollution problems caused by the coupling fluid, keeps the inspection site clean, and reduces the impact on the surrounding environment. The elimination of the need for applying and cleaning the coupling fluid significantly simplifies the inspection process, saves manpower and time costs, and improves inspection efficiency.

[0043] In this embodiment, the visual inspection module 13 includes a high-resolution camera mounted on the central frame 20. When scanning the weld, it can capture images of the weld surface at a certain frame rate and scanning path, obtaining image data containing rich details such as the overall shape and texture of the weld. This provides accurate data support for subsequent defect identification. After the high-resolution camera acquires the weld image, algorithms are used to analyze and process the image, automatically identifying common surface defects such as undercut, weld beads, and incomplete welds. The visual probe's appearance inspection module can not only detect surface defects but also compare the location information of these defects with internal defect data obtained from ultrasonic testing. Through techniques such as coordinate matching, the correlation between surface defects and internal defects is determined. All inspection data is integrated to output a comprehensive inspection result. This provides technicians with a comprehensive and accurate weld quality status, facilitating timely implementation of appropriate repair measures to ensure the quality and safety of the weld.

[0044] Example 2

[0045] This embodiment proposes a method for weld defect detection using the inspection robot described in Embodiment 1, including the following steps:

[0046] S1. The robot starts and positions the weld seam using its highly adaptable mobile platform. The omnidirectional movement of the mobile platform makes it extremely flexible. It can freely adjust its position in confined spaces and quickly align with the target weld seam without significant adjustments to its body direction. Combined with the robot's onboard positioning sensors, it can acquire its own position information in real time. Through precise motion control algorithms, it follows a preset path or adjusts its movement trajectory according to the actual environment to accurately stop at the weld seam that needs to be inspected.

[0047] S2. The robot's positioning sensors capture the probe's movement trajectory, while an ultrasonic detection module performs inspections and processes the ultrasonic time and frequency domain signals. A visual inspection module is used to detect defects and extract surface texture features. After acquiring the weld surface image, image processing algorithms are applied to preprocess the image to improve image quality and feature recognizability. Then, a texture analysis algorithm is used to extract the texture features of the defect surface image, serving as an important basis for subsequent defect identification and classification.

[0048] S3. The processed ultrasonic signal features are fused with the extracted visual texture features, and the three-dimensional morphology of the weld defect is obtained through point cloud reconstruction. Finally, a model that accurately reflects the three-dimensional morphology of the weld defect is obtained, providing intuitive and accurate geometric information for subsequent defect detection and classification.

[0049] S4. Defect detection and classification of the three-dimensional shape are performed using deep learning algorithms.

[0050] S5. Generate a diagnostic report.

[0051] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations; any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the protection scope of the present invention.

Claims

1. A weld defect inspection robot using a combination of ultrasonic and visual imaging, characterized in that: It includes a highly adaptable Mackerel moving platform and a weld defect detection unit located at the bottom of the highly adaptable Mackerel moving platform; The highly adaptable Mecanum wheel mobile platform includes a rear frame, a middle frame, and a Mecanum wheel assembly. The Mecanum wheel assembly includes a motor frame, a magnetic unit located at the bottom of the motor frame, a Mecanum wheel drive motor fixedly installed inside the motor frame, and Mecanum wheels driven to rotate by the Mecanum wheel drive motor. The rear frame is connected to the front and rear sides of the middle frame in a pitch-adjustable manner. The motor frame of the Mecanum wheel assembly is connected to the left and right sides of the rear frame in a position-adjustable manner. The weld defect detection unit includes an ultrasonic inspection module and a visual inspection module installed on the central frame.

2. The ultrasonic combined vision weld defect detection robot according to claim 1, characterized in that: The motor frame of the wheat wheel device is connected to the rear frame through a quadrilateral linkage mechanism. The rear frame is equipped with a wheat wheel adjustment mechanism for driving the quadrilateral linkage mechanism to deform and change the position of the wheat wheel device.

3. The ultrasonic combined vision weld defect detection robot according to claim 1, characterized in that: The rear frame and the middle frame are connected by several sets of upper and lower connecting rods. One end of the lower connecting rod is hinged to the middle frame, and the other end is slidably fitted into the strip groove of the middle frame. Both ends of the upper connecting rod are slidably fitted into the strip grooves of the rear frame and the middle frame, respectively. A spring is provided in the strip groove to spring against the sliding end of the connecting rod.

4. The ultrasonic combined vision weld defect detection robot according to claim 2, characterized in that: The rear frame includes a base plate and adjustment frames hinged to both sides of the base plate; several connecting rods are hinged between the motor frame and the adjustment frames, so that the adjustment frames, connecting rods and motor frame together form a quadrilateral linkage mechanism.

5. The ultrasonic combined vision weld defect detection robot according to claim 4, characterized in that: The wheel adjustment mechanism includes a screw and nut pair mounted on the rear frame, a slider driven by the nut to move horizontally, and a quadrilateral connecting frame. One side of the quadrilateral connecting frame is hinged to the slider, and the other side is hinged to the adjustment frame and fixedly connected to the connecting rod. Driven by the screw and nut pair, the slider drives the connecting rod to swing through the quadrilateral connecting frame, causing the quadrilateral connecting rod mechanism to deform and complete the position adjustment of the wheel device.

6. The ultrasonic combined vision weld defect detection robot according to claim 1, characterized in that: The ultrasonic testing module includes a probe shaft, an ultrasonic probe, a probe housing, and a probe outer wheel; the probe shaft is horizontally mounted at the bottom of the central frame, and several ultrasonic probes are distributed on the probe shaft to form an ultrasonic probe array; the probe housing is mounted on the probe shaft via bearings and surrounds the ultrasonic probes, and the two sides of the probe housing are sealed with sealing rings and filled with coupling fluid; the probe outer wheel is fixedly fitted onto the probe housing.

7. The ultrasonic combined vision weld defect detection robot according to claim 1, characterized in that: The motor frame is supported by a top plate via a shock absorber. The lower end of the shock absorber is hinged to the motor frame, and the upper end is hinged to a shock absorber mounting base at the bottom of the top plate.

8. The ultrasonic combined vision weld defect detection robot according to claim 7, characterized in that: A knob frame is mounted on the bottom of the top plate via a bracket, and the knob frame is equipped with an adjustment knob fixed to the upper end of the lead screw.

9. The ultrasonic combined vision weld defect detection robot according to claim 1, characterized in that: The vision inspection module includes a high-resolution camera mounted on the central frame.

10. A method for detecting weld defects using the inspection robot according to any one of claims 1-9, characterized in that, Includes the following steps: S1. The robot starts and positions the weld seam using a highly adaptable mobile platform; S2. Capture the probe's motion trajectory, simultaneously perform detection using the ultrasonic testing module, and process the ultrasonic time-domain and frequency-domain signals; perform detection using the visual testing module, and extract the texture features of the defect surface image; S3. The processed ultrasonic signal features are fused with the extracted visual texture features, and the three-dimensional morphology of the weld defect is obtained by point cloud reconstruction. S4. Defect detection and classification of the three-dimensional morphology are performed using deep learning algorithms; S5. Generate a diagnostic report.

Images