CN220764376U - Railway sleeper rail nut looseness detection robot based on machine vision - Google Patents

Abstract

The utility model discloses a railway sleeper nut loosening detection robot based on machine vision, which comprises a frame and mechanical arms fixedly arranged on two sides of the frame, wherein the tail ends of the mechanical arms are fixedly provided with image collectors; the front end and the rear end of the frame are respectively and fixedly provided with a group of guide rods, the bottom ends of the guide rods are provided with guide wheels capable of rolling along the side surfaces of the rails, the lower surface of the frame is fixedly provided with a group of T-shaped brackets, the inner sides of the T-shaped brackets are fixedly provided with driving mechanisms, the driving mechanisms are in transmission connection with first travelling wheels, and the two first travelling wheels are arranged in parallel; the automatic sleeper nut loosening detection device is characterized in that an image acquisition integrated circuit board is fixedly arranged on the frame, the image acquisition integrated circuit board is electrically connected with an image acquisition controller, a power supply and a remote control switch receiving device, and a driving mechanism, a mechanical arm and the image acquisition device are electrically connected with the power supply and the image acquisition controller, can automatically walk on rails and automatically guide, and is more suitable for automatic sleeper nut loosening detection under a complex environment.

Description

Railway sleeper rail nut looseness detection robot based on machine vision

Technical Field

The utility model relates to rail detection equipment, in particular to a railway sleeper nut loosening detection robot based on machine vision.

Background

The nuts on the railway sleeper are connected with the bolts, so that reliable connection between the steel rail and the sleeper can be effectively ensured for a long time, the buffering and damping effects are exerted under the action of power, but the sleeper nuts can loose under the influence of long-time vibration and natural factors due to train operation. Therefore, ensuring the tightness of the nuts of the railway sleeper to meet the standard is an important problem for the operation safety of the related railway.

At present, in the railway sleeper nut tightness detection method, manual detection is most common, but the manual detection seriously depends on human senses, has the defects of poor working conditions, high labor intensity and low efficiency, is greatly influenced by subjective factors of operators, and is screwed once under the condition that maintenance workers are forced to frequently judge in order to ensure safety. Thus, the workload of maintenance workers is increased, and the service life of the nut is also reduced. In addition, there are detection based on structural vibrations, detection based on sound collection analysis, and image detection based on machine vision, wherein: detecting the loosening condition of the bolt by extracting the change of characteristic parameters such as characteristic frequency, transfer function, power spectrum and the like of the structure before and after loosening of the bolt based on the detection of the structure vibration, but the initial loosening of the nut cannot be effectively detected; the detection based on sound collection analysis utilizes an electromagnet to excite the bolt with equal amplitude and equal frequency, after sound information is collected through a microphone module, the abnormal loosening characteristics are found out through analysis, and loosening detection is realized. The image detection method based on machine vision mainly comprises information detection, acquisition systems, image processing, display, intelligent decision making and other modules, is mainly carried out in a laboratory environment, and establishes a nut loosening dataset which is an important basis for subsequently improving bolt image detection precision, but only manually acquires bolt data, so that the working strength is high, and unified standard cannot be established, therefore, a sleeper nut loosening detection robot for acquiring sleeper nut loosening data images is designed by a learner.

The existing railway sleeper nut loosening detection robot mainly comprises a traveling mechanism, a frame and an acquisition system retraction device, wherein the traveling mechanism is an important part of the railway sleeper nut loosening detection robot, traveling is unstable or a complex road cannot smoothly pass through, accuracy of detection results can be influenced, the traveling mechanism applicable to the railway sleeper nut loosening detection robot at present is mainly a crab-like single-rail moving device, four right-angle guide wheels of the traveling mechanism are fixedly arranged at the bottom of a frame respectively and are arranged in a two-to-two straight line mode and are used for being in rolling fit with two side edges of the upper end of a rail, an output end of a driving motor is connected with a driving wheel through a transmission device, the driving wheel is used for being in rolling fit with the upper end face of the rail, the device is enabled to move, however, a groove part of the V-shaped wheel is in contact with the side edges of the rail and rolls along the side edges of the rail, and simultaneously plays roles of traveling and guiding, however, according to the characteristics of the double-rail train traveling mechanism, two sides of a single rail are different in smoothness, namely, the side faces of the rail are not smooth, the rail side faces are easy to wear when the V-shaped wheels are severely, the body is easy to vibrate, the structure is enabled to vibrate, and the detection is not accurate, and service life is shortened. In addition, the bolt operational environment is more in the field, and the nut periphery is easily sheltered from by weeds, and current detection robot shooting scope is less, consequently hardly beats clear image, and then can't accurately judge the nut loosening condition.

Disclosure of Invention

Aiming at the defects existing in the prior art, the utility model aims to provide a railway sleeper nut loosening detection robot based on machine vision, which can stably walk on a rail and automatically guide, thereby improving the detection precision and prolonging the service life.

In order to achieve the above purpose, the utility model is realized by adopting the following technical scheme:

a railway sleeper nut loosening detection robot based on machine vision comprises a frame and mechanical arms fixedly arranged on two sides of the frame, wherein an image collector is fixedly arranged at the tail end of each mechanical arm;

the front end and the rear end of the frame are respectively and fixedly provided with a group of guide rods, the bottom ends of the guide rods are provided with guide wheels capable of rolling along the side surfaces of the rails, the lower surface of the frame is fixedly provided with a group of T-shaped brackets, the inner sides of the T-shaped brackets are fixedly provided with driving mechanisms, the driving mechanisms are in transmission connection with first travelling wheels, and the two first travelling wheels are arranged in parallel;

the image acquisition integrated circuit board is fixedly arranged on the frame and electrically connected with an image acquisition controller, a power supply and a remote control switch receiving device, and the driving mechanism, the mechanical arm and the image acquisition device are electrically connected with the power supply and the image acquisition controller.

Further, the mechanical arm comprises a base fixedly connected with the frame, the base is connected with a waist through a waist joint, the waist is connected with a big arm through a shoulder joint, the big arm is connected with a small arm through an elbow joint, and the tail end of the small arm is fixedly connected with the image collector;

the waist joint, the shoulder joint and the elbow joint comprise steering engines and steering wheels fixedly connected with output shafts of the steering engines.

Further, the big arm and the small arm comprise two hard aluminum plates, and the two hard aluminum plates are fixedly connected through a U-shaped aluminum plate.

Further, the driving mechanism comprises a motor, an output shaft of the motor is fixedly connected with a worm, the worm is connected with a worm wheel in a transmission manner, and the equivalent friction angle of the worm is smaller than the lead angle;

an output shaft of the worm gear is fixedly connected with the first traveling wheel.

Further, the lower surface of the frame is fixedly provided with a fork-shaped bracket, and the fork-shaped bracket is rotationally connected with a second travelling wheel.

Further, concave guide rod supporting frames are fixedly arranged at the front end and the rear end of the frame, and the guide rods are fixedly connected with the guide rod supporting frames.

Further, the mechanical arm support frames are fixedly arranged on two sides of the frame, and the base is fixedly connected with the mechanical arm support frames.

Further, the guide wheel is a rubber wheel.

Further, the diameter of the guide wheel is 24mm, 21mm or 32mm.

Further, the first travelling wheel and the second travelling wheel are rubber wheels or nylon wheels.

Compared with the prior art, the utility model has the following technical effects:

the utility model adopts a monorail straddle type structure, the walking and the guide are independent and mutually coordinated, namely, the guide is realized by arranging two groups of guide wheels capable of rolling along the side surfaces of the rail, the parallel first walking wheels are assisted to walk on the top surfaces of the rail, not only can the robot be prevented from falling off the rail, but also the robot can be assisted to stably turn, the parallel wheels and the guide wheels work in a coordinated manner, the robot can more stably advance and retreat, the machine body is not easy to shake, so that the detection accuracy is improved, in addition, the parallel wheels and the rail top surfaces which are relatively smooth in the rail are in surface contact, the abrasion of the walking wheels is reduced, the service life of the walking wheels is prolonged, the stability is good, the detection accuracy is further improved, the robot can completely replace manual detection, and the working strength of workers is reduced.

The three-degree-of-freedom mechanical arm is adopted, the space movable range is larger, the shooting angle of the image collector is more flexible and changeable, even in overcast and rainy days or under complex environments such as being blocked by sundries, the steering engine is driven by the image collection controller to finely adjust the mechanical arm according to the image collection requirement, so that the image with clear, usable and high resolution is collected, the detection of the loosening condition of the sleeper nut is automatically completed, the environment factor is not limited, the loosening condition of the nut can be accurately detected, and the automatic detection device is suitable for the automatic detection of the loosening condition of the rail nut under the complex environments.

In the process of finishing advancing, retreating, parking and turning along the track, the robot can not only advance or retreat at a constant speed, but also slow down, so that the acquisition process is more convenient; in addition, the second walking wheel is a single wheel, and the robot can play a role in balancing and stabilizing in the walking process.

Adopt the leading wheel of rubber material and the walking wheel of rubber or nylon material, can play the cushioning effect, no longer need additionally set up damping device, make the overall structure of robot simpler.

Drawings

Fig. 1: the sleeper nut loosening detection flow chart is provided;

fig. 2: the structure of the utility model is schematically shown in the use state;

fig. 3: the side view structure schematic diagram of the utility model in the use state;

fig. 4: the front view structure schematic diagram of the utility model;

fig. 5: the partial structure schematic diagram of the mechanical arm is provided;

fig. 6: the first travelling wheel is structurally schematic;

fig. 7: the driving mechanism of the utility model is structurally schematic;

fig. 8: the structure of the first travelling wheel of the driving mechanism is schematically shown;

fig. 9: the structure of the guide mechanism is schematically shown;

fig. 10: the structure of the frame is schematically shown;

fig. 11: the structure of the guide wheel is schematically shown;

fig. 12: the guide wheel structure schematic diagram with the wheel diameter of 24mm is provided;

fig. 13: the structure of the guide wheel with the diameter of 21mm is schematically shown;

fig. 14: the structure of the guide wheel with the wheel diameter of 32mm is schematically shown;

fig. 15: the coordinate system of the mechanical arm is provided by the utility model;

fig. 16: the mechanical arm is structurally schematic;

in the figure: 1. a mechanical arm bracket; 2. a base; 3. a waist portion; 4. a guide rod; 5. a guide wheel; 6. a large arm; 7. steering engine; 8. an image collector; 9. a guide rod support; 10. a frame; 11. an image acquisition controller; 12. a power supply; 13. an image acquisition integrated circuit board; 14. a remote control switch receiving device; 15. a fork-shaped bracket; 16. a first traveling wheel; 17. a driving mechanism; 18. a T-shaped bracket; 19. a forearm; 20. steering wheel; 21. a second travelling wheel; 22. a motor; 23. a worm wheel; 24. a worm; 25. a rail.

Detailed Description

The following examples illustrate the utility model in further detail.

As shown in fig. 2-16, a machine vision-based railway sleeper nut loosening detection robot comprises a frame 10, a running mechanism, a guide device, an image acquisition mechanism and a control system, wherein the running mechanism, the guide device, the image acquisition mechanism and the control system are connected with the frame;

the structure of the frame 10 should meet the requirements of the overall arrangement of the robot: firstly, when the robot runs in a complex and changeable way, all assemblies and parts fixed on the frame 10 should not interfere; secondly, when the robot runs on the track, the frame 10 may generate torsion deformation and bending deformation in the longitudinal plane under the action of load, and the deformation will change the relative positions of the components mounted on the frame 10, thereby affecting the normal operation of the robot, therefore, the frame 10 should have enough strength, and the mass is as small as possible to reduce the whole mass of the robot, in addition, the height of the frame 10 is reduced as much as possible, so that the central position of the robot is reduced, and the running stability of the robot is improved;

according to the standard head widths (60 mm, 68mm, 70mm, 73mm and 75 mm) of heavy rails (YB 350-63, GB183-63, GB182-63 and GB181-63) in China, the length of the designed rack 10 is 260mm, the width is 160mm and the distance from the track surface is 70mm by taking the head width of 75mm as the standard.

As shown in fig. 2-4, fig. 9 and fig. 10, the guide device comprises guide rod supporting frames 9 fixedly installed at the front end and the rear end of the frame 10 through bolts, the guide wheel supporting frames 9 are set to be 170mm long and 10mm wide according to the size of the frame 10, vertical guide rods 4 are fixedly installed at the two ends of the guide rod supporting frames 9, guide wheels 5 capable of rolling along the side surfaces of the rails 25 are installed at the bottom ends of the guide rods 4, the two groups of guide wheels 5 clamp the left side and the right side of the rails 25, the robot can be prevented from falling out of the rails 25, the guide device can assist the robot to complete turning when encountering a curve, the guide wheels 5 are required to have conditions of wear resistance, shock absorption and the like, therefore screw rubber-covered bearings are selected, according to the heavy rail standard in China, the minimum passing radius of the turning position of the rails 25 is 100m, and when encountering the curve, the length of the robot is generally smaller than the radius of the curve, and the robot can properly turn along with the rails 25 under the guidance of the guide device.

As shown in fig. 11-14, according to different track type head widths, different sizes of guide wheels 5 with wheel diameters of 24mm, 21mm, 32mm and the like can be selected, and in this embodiment, the head width of the detection track is 75mm, so that the guide wheel 5 with the wheel diameter of 24mm is selected, and in order to increase the stability and shock absorption performance of the guide wheel 5 during guiding, the guide wheel 5 is a rubber wheel.

As shown in fig. 2-4 and fig. 6-8, the travelling mechanism comprises a group of t-shaped brackets 18 fixedly installed on the lower surface of the frame 10, a driving mechanism 17 is fixedly installed on the inner side of each t-shaped bracket 18, each driving mechanism 17 comprises a motor 22, each motor 22 adopts a small direct current motor, an output shaft of each motor 22 is fixedly connected with a worm 24, the worm 24 is in transmission connection with a worm wheel 23, an output shaft of each worm wheel 23 is fixedly connected with a first travelling wheel 16, and the two first travelling wheels 16 are arranged side by side to form a parallel travelling wheel; the lower surface of the frame 10 is fixedly provided with a fork-shaped bracket 15, the fork-shaped bracket 15 is rotationally connected with a second travelling wheel 21, the second travelling wheel 21 is a single wheel and is arranged right behind the parallel wheels, and the first travelling wheel 16 and the second travelling wheel 21 are all rubber wheels or nylon wheels, so that the whole structure of the robot is simple, and the damping effect is improved;

in addition, according to the design requirement that the robot can move forwards and backwards, the self-locking function of the worm and gear driving mechanism is eliminated in the worm and gear driving mechanism, namely the equivalent friction angle of the worm is smaller than the lead angle.

As shown in fig. 2-5 and fig. 15-16, the image acquisition mechanism comprises a mechanical arm support frame 1 fixedly installed on two sides of a frame 10 and a mechanical arm fixedly connected with the mechanical arm support frame, the mechanical arm comprises a base 2 fixedly connected with the mechanical arm support frame 1, the base 2 is connected with a waist 3 through a waist joint, the waist 3 is connected with a big arm 6 through a shoulder joint, the big arm 6 is connected with a small arm 19 through an elbow joint, the tail end of the small arm 19 is fixedly connected with an image acquisition device 8, when a robot walks to a sleeper nut position, the mechanical arm can automatically adjust the shooting angle of the image acquisition device 8, the two image acquisition devices 8 are utilized to respectively acquire and record sleeper nut states on two sides of a rail 25 and then the sleeper nut states are transmitted to a control system, and the waist joint, the shoulder joint and the elbow joint comprise a steering engine 7 and a steering wheel 20 fixedly connected with an output shaft of the steering wheel;

because the actual sleeper nut working environment is complex, for example, weed shielding exists on the periphery, in order to acquire the most clear image, as shown in fig. 15, the waist joint can rotate around the z axis, the rotation range is 0-180 degrees, the shoulder joint and the elbow joint rotate around the x axis, in order to ensure that the camera is parallel to the bolt axis, the rotation range is set to be 0-150 degrees (namely, the small arm 19 moves in a pitching way within the range of 0-150 degrees), the pitching actions of the large arm 6 and the small arm 19 jointly determine the position of the image acquisition device 8 in the horizontal plane, the large arm 6 and the small arm 19 are of hollow structures, the steering engine 7 is arranged at the joint of the large arm to control the rotation of the large arm, and the rotation of the waist 3 and the pitching actions of the large arm 6 and the small arm 19 jointly determine the spatial position of the image acquisition device 8; the waist 3 is in a hollowed-out design, the waist 3 is rotated by transmitting torque through the steering engine 7, the waist 3 is connected with the steering engine 7 through the steering wheel 20, and the steering engine 7 drives the waist 3 to rotate, so that the mechanical arm rotates;

preferably, the arm lengths of the large arm 6 and the small arm 19 are 90mm.

Preferably, in order to lighten dead weight, the big arm 6 and the small arm 19 all comprise two hard aluminum plates, the two hard aluminum plates are fixedly connected through a U-shaped aluminum plate, a clamping groove for installing the steering engine 7 is reserved between the two hard aluminum plates forming the small arm 19, and the steering engine 7 outputs from a joint of the small arm 19, so that the image collector 8 is driven to adjust the angle.

As shown in fig. 2, the control system includes an image acquisition integrated circuit board 13 fixedly installed on the frame 10, the image acquisition integrated circuit board 13 is electrically connected with an image acquisition controller 11, a power supply 12 and a remote switch receiving device 14, and the motor 22, the mechanical arm and the image acquisition device 8 are electrically connected with the power supply 12 and the image acquisition controller 11.

The working principle of the utility model is as follows:

as shown in fig. 1, during detection, the image acquisition controller 11 sends out a signal and turns on the motor 22, the first travelling wheel 16 is driven to roll along the top surface of the rail 25 so as to drive the robot to move forward, the guide wheel 5 clamps the left side and the right side of the rail 25, the robot can be prevented from falling out of the rail 25, when the robot encounters a curve, the guide wheel 5 can assist the robot to complete turning, when the image acquisition device 8 needs to acquire a photo, the image acquisition controller 11 sends out a signal and turns off the motor 22 so as to stop the robot, due to the influence of ascending and descending slopes, the actual stop position of the robot possibly exceeds the view of the image acquisition device 8 for shooting a nut, the image acquisition controller 11 sends out a signal again and drives the motor 22 to reverse so as to enable the robot to retreat, when the nut completely enters the shooting view of the image acquisition device 8, the robot stops again, the image acquisition controller 11 sends out a signal and drives the mechanical arm to rotate and pitch so as to adjust the spatial position of the image acquisition device 8, so as to shoot clear real-time images, and transmit image data to the image acquisition controller 11, and the image acquisition controller 11 compares the acquired data with the data integrated in advance so as to judge whether the nut is loose or not.

Claims (10)

1. The railway sleeper nut loosening detection robot based on machine vision is characterized by comprising a frame (10) and mechanical arms fixedly arranged on two sides of the frame, wherein an image collector (8) is fixedly arranged at the tail end of each mechanical arm;

a group of guide rods (4) are fixedly arranged at the front end and the rear end of the frame (10), guide wheels (5) capable of rolling along the side surfaces of the rails (25) are arranged at the bottom ends of the guide rods (4), a group of T-shaped brackets (18) are fixedly arranged on the lower surface of the frame (10), a driving mechanism (17) is fixedly arranged on the inner side of each T-shaped bracket (18), a first travelling wheel (16) is connected with the driving mechanism (17) in a transmission manner, and the two first travelling wheels (16) are arranged in parallel;

an image acquisition integrated circuit board (13) is fixedly arranged on the frame (10), the image acquisition integrated circuit board (13) is electrically connected with an image acquisition controller (11), a power supply (12) and a remote control switch receiving device (14), and a driving mechanism (17), a mechanical arm and the image acquisition device (8) are electrically connected with the power supply (12) and the image acquisition controller (11).

2. The machine vision-based railway sleeper nut looseness detection robot according to claim 1, wherein the mechanical arm comprises a base (2) fixedly connected with a frame (10), the base (2) is connected with a waist (3) through a waist joint, the waist (3) is connected with a big arm (6) through a shoulder joint, the big arm (6) is connected with a small arm (19) through an elbow joint, and the tail end of the small arm (19) is fixedly connected with an image collector (8);

the waist joint, the shoulder joint and the elbow joint comprise steering engines (7) and rudder discs (20) fixedly connected with output shafts of the steering engines.

3. The machine vision-based railway sleeper nut loosening detection robot according to claim 2, wherein the large arm (6) and the small arm (19) each comprise two hard aluminum plates, and the two hard aluminum plates are fixedly connected through a U-shaped aluminum plate.

4. A machine vision based railway sleeper nut loosening detection robot as claimed in any one of claims 1-3, characterized in that the drive mechanism (17) comprises a motor (22), an output shaft of the motor (22) is fixedly connected with a worm (24), the worm (24) is in transmission connection with a worm wheel (23), and an equivalent friction angle of the worm (24) is smaller than a lead angle;

the output shaft of the worm wheel (23) is fixedly connected with the first travelling wheel (16).

5. The machine vision-based railway sleeper nut loosening detection robot as claimed in any one of claims 1 to 3, wherein a fork-shaped bracket (15) is fixedly installed on the lower surface of the frame (10), and the fork-shaped bracket (15) is rotatably connected with a second travelling wheel (21).

6. The machine vision-based railway sleeper nut loosening detection robot according to any one of claims 1 to 3, wherein concave guide rod supporting frames (9) are fixedly installed at the front end and the rear end of the frame (10), and the guide rods (4) are fixedly connected with the guide rod supporting frames (9).

7. The machine vision-based railway sleeper nut loosening detection robot according to claim 2 or 3, wherein the mechanical arm support frames (1) are fixedly installed on two sides of the frame (10), and the base (2) is fixedly connected with the mechanical arm support frames (1).

8. A machine vision based railway sleeper nut loosening detection robot as claimed in any one of claims 1-3, wherein the guide wheel (5) is a rubber wheel.

9. A machine vision based railway sleeper nut looseness detection robot according to any of claims 1-3, wherein the wheel diameter of the guide wheel (5) is 24mm, 21mm or 32mm.

10. A machine vision based railway sleeper nut looseness detection robot in accordance with any of claims 1-3, wherein the first and second travelling wheels (16, 21) are rubber or nylon wheels.