Intelligent track flaw detection equipment and flaw detection method

Abstract

The invention provides intelligent track flaw detection equipment and a flaw detection method, and belongs to the technical field of track flaw detection. According to the method, an innovative strategy of staged flaw detection is adopted, double-mode dynamic switching of rapid rough flaw detection and low-speed fine flaw detection is utilized, a high-frame-rate visual system is used for screening surface defects, ultrasonic wave and laser ranging are combined, deep cracks and track gauge deviation are detected, all defect types are covered, and defect missing detection errors at a single speed are compensated. The flaw detection equipment comprises a flaw detection trolley, the flaw detection trolley comprises a chassis, and the chassis is composed of an advancing cross beam and two groups of advancing longitudinal beams connected to the end parts of the advancing cross beam; the advancing cross beam is connected with a visual system and a laser range finder, and an ultrasonic flaw detector is arranged on the advancing longitudinal beam. According to the invention, ultrasonic wave, machine vision and laser ranging technologies are integrated, full-dimension, full-climate and full-scene track flaw detection is realized, and the railway operation and maintenance efficiency, the intelligent level and the operation safety are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of track flaw detection, and specifically relates to an intelligent track flaw detection device and a flaw detection method. Background Art

[0002] For track flaw detection equipment, hand-pushed single-rail ultrasonic flaw detectors are widely used in China at present. This equipment has the characteristics of small volume and flexible use, but it can only perform single-rail detection, with low efficiency. When the coupling condition is not good, problems such as wave loss and missed detection are likely to occur, and there is still room for improvement in terms of accuracy and precision.

[0003] Currently, foreign equipment (SPERRY in the United States, GE in Germany) widely uses ultrasonic flaw detection technology to detect internal defects of tracks using high-frequency sound waves. It mostly relies on single-frequency ultrasonic detection, with limited detection depth (usually ≤ 100 mm), and poor adaptability to complex environments (rain, snow, high temperature). Some foreign equipment (the AI flaw detection system of the Japanese Shinkansen) introduces machine vision technology to capture surface defects of tracks through cameras. However, its false positive rate is still relatively high (the system false positive rate is about 8%), and manual intervention is required at night or under low light conditions. Laser ranging technology is used for gauge measurement, but most foreign equipment uses a single laser ranging method, lacking the ability of multi-source data fusion, resulting in limited measurement accuracy and efficiency.

[0004] In summary, through research on existing publicly reported track flaw detection equipment at home and abroad, it is found that most of them have problems such as a single technical route, low detection efficiency, poor environmental adaptability, lack of multi-modal data fusion, and a high fault misjudgment rate.

[0005] With the rapid development of high-speed railways, railway equipment operates at high load, and potential hazards such as rail fatigue and gauge deformation occur frequently, greatly increasing the demand for track flaw detection. However, the traditional single method can no longer meet the new requirements. Based on this, a "track" detective doctor has been developed to meet the timeliness of railway maintenance. Summary of the Invention

[0006] Based on the problems existing in the background art, the present invention proposes an intelligent track flaw detection device and a flaw detection method. The present invention integrates ultrasonic, machine vision, and laser ranging technologies, can detect track defects in all directions, solve pain points such as complex operation, and greatly improve railway operation and maintenance efficiency, intelligent level, and operation safety. The present invention focuses on solving safety hazards such as surface and deep cracks of tracks, gauge deformation, and sleeper deterioration. By integrating three core technologies of Duffing chaotic oscillator ultrasonic detection, machine vision, and laser dynamic ranging, "full-dimensional, all-weather, and full-scenario" track flaw detection is realized, significantly improving detection efficiency and accuracy, reducing manual dependence, and providing an intelligent maintenance solution for railways, subways, and heavy-haul lines.

[0007] The technical solution is as follows: In a first aspect, the present technical solution proposes an intelligent rail flaw detection method, which at least includes a fast rough flaw detection mode. The fast rough flaw detection mode includes machine vision detection and laser gauge measurement, where: The method of machine vision detection is as follows: Image acquisition: Capture the surface image of the rail through the vision system on the flaw detection trolley; Image preprocessing: Obtain the preprocessed rail image through noise reduction, enhancement, and segmentation; Defect analysis and classification: Input the preprocessed rail image into the YOLOV10 model. After in-depth analysis and processing, determine the type of surface defect and obtain the rail surface defect detection result; The method of laser gauge measurement is as follows: Gauge data acquisition: Synchronously collect gauge data through the laser rangefinder on the flaw detection trolley and mark the gauge over-limit area in real time; Data processing: Based on the image data after image preprocessing in machine vision detection and the gauge data in laser gauge measurement, fuse them through the edge computing module to generate a heat map of suspicious areas, and display the defect results and mark the defects.

[0008] Preferably, it further includes a low-speed fine flaw detection mode, and the specific method is as follows: Ultrasonic data acquisition and preprocessing: Detect the defects inside the rail through the ultrasonic flaw detector on the flaw detection trolley, collect the ultrasonic data and perform preprocessing to obtain the detection result of deep rail defects; Data processing: Input the image data after image preprocessing in machine vision detection and the ultrasonic data into the federated learning framework to dynamically optimize the defect classification model for processing, determine the defect type and classify it, generate the final rail defect detection result and visualize the output.

[0009] Preferably, in the ultrasonic data acquisition step, the Duffing chaotic oscillator subsystem is used to process the ultrasonic data, and the specific method is as follows: Signal acquisition: Use the ultrasonic probe of the ultrasonic flaw detector to obtain the echo signal; Chaotic system drive: Input the obtained echo signal into the Duffing system to reduce the noise interference of ultrasonic detection; Signal amplification and detection: The Duffing system detects the echo signal after noise reduction processing, and at the same time extracts and amplifies the characteristics of the detected weak signal to obtain the processed echo signal data; Data analysis and processing: Analyze the processed echo signal data to determine the defect information.

[0010] Preferably, the vision system uses a CMOS vision system.

[0011] Preferably, the defect types include, but are not limited to, cracks, wear, rust, and weld defects.

[0012] In a second aspect, the present technical solution provides an intelligent rail flaw detection device for performing the intelligent rail flaw detection method described above. The intelligent rail flaw detection device includes a flaw detection trolley, which includes a chassis composed of a traveling cross beam and two sets of traveling longitudinal beams connected to the ends of the traveling cross beam; a vision system and a laser rangefinder are connected to the traveling cross beam, and an ultrasonic flaw detector is provided on the traveling longitudinal beams; A liquid storage tank, detachably connected above the traveling longitudinal beams, and a coupling agent inside the liquid storage tank is coated above the rail through a spraying assembly; A traveling system, arranged on the chassis, for driving the flaw detection trolley to move.

[0013] Preferably, it further includes two sets of dialing and discharging assemblies, both arranged at the bottom of the traveling longitudinal beams, which alternately wipe the liquid above the rail. The alternating dialing and discharging assembly includes a driving motor arranged inside the traveling longitudinal beam. The output shaft at the bottom of the driving motor is connected to a swing arm, and a dialing plate is hinged to the bottom of the swing arm; it also includes an auxiliary rod hinged to the bottom of the traveling longitudinal beam, and the other end of the auxiliary rod is rotatably connected to a dialing plate; when the driving motor drives the swing arm to rotate, it can drive the dialing plate to reciprocate and dial outwards.

[0014] Preferably, a supporting fixture is fixedly connected above the traveling longitudinal beam, and the liquid storage tank is placed above the supporting fixture. The liquid storage tank is connected to the spraying assembly through a liquid discharge channel at the bottom of the supporting fixture.

[0015] Preferably, the supporting fixture has a rectangular structure. Grooves are provided on the four side walls of the supporting fixture, and limiting blocks are hinged in the grooves. The limiting blocks have an L-shaped structure, and a clamping block is fixedly connected to the top of the limiting blocks; a clamping groove that can be clamped with the clamping block is provided on the outer wall of the liquid storage tank; after the liquid storage tank is placed in the supporting fixture, it presses the bottom of the limiting block, so that the clamping block is clamped with the clamping groove, positioning the liquid storage tank above the supporting fixture.

[0016] The above technical solution has the following advantages: In terms of the method: 1. Aiming at the problem of single rail detection equipment: integrating ultrasonic flaw detection, machine vision recognition, and gauge measurement technologies, solving the problems of single function, poor environmental adaptability, and missed detection of traditional equipment. The supporting intelligent data analysis system realizes the synchronous and accurate identification of surface and deep defects of the rail, with high detection accuracy, breaking through key technologies such as ultrasonic signal amplification, visual dynamic recognition, and laser gauge dynamic measurement, providing core technical support for the safe operation and intelligent upgrade of railways.

[0017] 2. To address the problem of low detection efficiency, an innovative flaw detection strategy is proposed: a phased flaw detection strategy. By dynamically switching between two modes of rapid rough flaw detection and low-speed fine flaw detection, a high-frame-rate vision system is used to screen for surface defects. Combining ultrasonic and laser ranging technologies, deep cracks and gauge deviations are detected, covering all types of defects. The error of missed detection at a single speed is compensated, and multiple flaw detection devices are rationally coordinated to achieve collaborative operation. A data management platform is built to centrally store, manage, and analyze flaw detection data, providing support for operations and effectively shortening the flaw detection time.

[0018] 3. To effectively solve the problems of data fusion and data accuracy: Key technologies such as multimodal data fusion, real-time noise filtering, and micro-defect classification have been broken through, significantly different from the traditional manual experience-based interpretation mode, enabling precise identification and dynamic optimization of defects under complex working conditions. The Duffing chaotic oscillator technology is innovatively utilized to increase the signal-to-noise ratio by 40%. The dynamic allocation of multi-sensor weights using a deep learning framework promotes the leap of detection technology from "manual experience" to "data intelligence".

[0019] On the device: The coupling agent on the surface of the rail is pushed to one side of the rail by a row-pushing component. Two groups of row-pushing components form a cycle of "wiping - resetting - wiping", achieving non-stop alternating operation, which can greatly improve the cleaning of the rail surface and the uniform coating of the coupling agent, creating better coupling conditions and reducing the incidence of wave loss and missed detection problems. Brief Description of the Drawings

[0020] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to actual scale.

[0021] Figure 1 It is a three-dimensional structural schematic diagram of the present invention.

[0022] Figure 2 It is a front view of the present invention.

[0023] Figure 3 It is a structural schematic diagram of the row-pushing component at the bottom of the traveling longitudinal beam in the present invention (the dashed lines respectively represent the internal structure of the traveling longitudinal beam and the track line).

[0024] Figure 4 It is a structural schematic diagram of the row-pushing component between the traveling longitudinal beam and the rail in the present invention.

[0025] Figure 5 It is a schematic diagram of the connection relationship between the supporting fixture and the liquid storage tank in the present invention.

[0026] Figure 6It is a cross-sectional view of the connection relationship between the liquid storage tank and the traveling longitudinal beam in the present invention.

[0027] Figure 7 It is a flowchart of the method of the present invention.

[0028] Figure 8 It is a schematic structural diagram of defect result display and defect marking. Figure 9 It is a three-dimensional view of the intelligent rail flaw detection device proposed by the present invention from another perspective. Figure 10 It is a schematic structural diagram of the intelligent rail flaw detection device proposed by the present invention during use.

[0029] Explanation of reference numerals: 1. Traveling cross beam; 101. Main body beam; 102. Connecting beam; 103. Connecting component; 2. Traveling longitudinal beam; 3. Vision system; 4. Ultrasonic flaw detector; 5. Liquid storage tank; 51. Card slot; 6. Spraying component; 61. Output pump; 62. Spraying box; 63. Nozzle; 7. Pushing and discharging component; 71. Driving motor; 72. Swing arm; 73. Pushing plate; 74. Auxiliary rod; 75. Trajectory line; 8. Recycling box; 9. Traveling wheel; 10. Recycling pump; 11. Recycling pipeline; 12. Recycling tank; 13. Supporting fixture; 14. Drainage channel; 15. Groove; 16. Limit block; 17. Clamping block; 18. Rail; 19. Laser rangefinder; 20. Auxiliary positioning wheel. Detailed implementation manners

[0030] Hereinafter, embodiments of the technical solution of the present invention will be described in detail with reference to the drawings. The following embodiments are only used to illustrate the technical solution of the present invention more clearly, so they are only examples and cannot be used to limit the protection scope of the present invention.

[0031] Embodiment 1: As Figure 7 - Figure 8 shown, this embodiment proposes an intelligent rail flaw detection method, including a fast and rough flaw detection mode. The fast and rough flaw detection mode includes machine vision detection and laser gauge measurement, which are elaborated in detail below: The method of machine vision detection generally includes several major steps such as image acquisition → image preprocessing → defect analysis and classification. Among them: Image acquisition: The vision system 3 on the flaw detection trolley is used to capture the surface image of the rail 18; in this embodiment, the vision system 3 can adopt a CMOS vision system 3.

[0032] Image preprocessing: The preprocessed image of the rail 18 is obtained through noise reduction, enhancement and segmentation.

[0033] Defect analysis and classification; the preprocessed rail 18 images are input into the YOLOV10 model. After in-depth analysis and processing, the surface defect type is determined to obtain the surface defect detection result of the rail 18.

[0034] The method of laser gauge measurement includes gauge data acquisition → data processing → defect result display and defect annotation, where: Gauge data acquisition: The gauge data is synchronously collected through the laser rangefinder 19 on the flaw detection trolley, and the gauge overlimit area is marked in real time; The gauge measurement technology based on laser ranging: This technology measures the track spacing in real time through high-precision laser dynamic ranging. Combined with the design of the drum-shaped pulley, it ensures the dynamic stability of the contact surface between the laser probe and the track, reduces the measurement error caused by vibration, and realizes high-precision dynamic measurement and full-automatic continuous detection.

[0035] Data processing: Based on the image data after image preprocessing in machine vision detection and the gauge data in laser gauge measurement as the data set, they are fused through the edge computing module to generate a heat map of suspicious areas, and defect result display and defect annotation are carried out. The defect result feedback is as Figure 8 shown.

[0036] The signal acquisition and image processing technology of the track surface based on machine vision: Through the collaborative innovation of the high-resolution CMOS vision system 3 and the YOLOV10 model, it can identify visible defects such as cracks, wear, and rust on the track surface, realizing the screening of surface defects; At the same time, the laser ranging technology is used to monitor the track spacing in real time to ensure compliance with safety standards.

[0037] In this embodiment, the intelligent track flaw detection method also includes a low-speed fine flaw detection mode, and the specific method is as follows: Ultrasonic data acquisition and preprocessing: The ultrasonic flaw detector 4 on the flaw detection trolley is used to detect the defects inside the track, and the ultrasonic data is collected and preprocessed to obtain the detection result of the deep-layer defects of the track; Data processing: The image data after image preprocessing in machine vision detection and the ultrasonic data are input into the federated learning framework to dynamically optimize the defect classification model for processing, determine the defect type and classify it, generate the final track defect detection result and visualize the output. The defect types include but are not limited to cracks, wear, rust, weld defects, etc. This method uses the federated learning framework to dynamically optimize the defect classification model, enhances the generalization ability of the defect classification model through distributed learning, and avoids the risk of data privacy.

[0038] In this embodiment, in the ultrasonic data acquisition step, the Duffing chaotic oscillator subsystem is used to process the ultrasonic data, and the specific method is as follows: Signal acquisition: Use the ultrasonic probe of the ultrasonic flaw detector 4 to obtain the echo signal; Chaotic system drive: Input the obtained echo signal into the Duffing system to reduce the noise interference in ultrasonic detection; Signal amplification and detection: The Duffing system detects the echo signal after noise reduction processing, and at the same time extracts and amplifies the features of the detected weak signal to obtain the processed echo signal data; Data analysis and processing: Analyze the processed echo signal data to determine the defect information.

[0039] In this method, the signal-to-noise ratio of weak ultrasonic signals is improved by the Duffing chaotic oscillator algorithm, and the recognition rate of deep defects is increased.

[0040] Internal ultrasonic detection technology for rail 18 based on the Duffing chaotic oscillator subsystem: Combine the nonlinear dynamic characteristics of the Duffing chaotic oscillator with ultrasonic flaw detection, utilize the chaotic edge signal amplification mechanism, take the ultrasonic echo as a perturbation to input into the system, and significantly improve the signal-to-noise ratio through state transition. The chaotic system naturally suppresses environmental noise, and the electromagnetic interference filtering rate ≥ 80%. The sensitivity of defect detection is increased by 56.3% and the detection accuracy is increased by 98.6%.

[0041] Beneficial effects: Aiming at the problem of low detection efficiency, this solution innovates the flaw detection strategy: Staged flaw detection strategy: Utilize the dynamic switching of dual modes of fast rough flaw detection and low-speed fine flaw detection. The high-frame-rate vision system 3 screens surface defects, combines ultrasonic waves and laser ranging to detect deep cracks and gauge deviations, covers all defect types, compensates for the missed detection error of defects at a single speed, rationally deploys various flaw detection devices to achieve collaborative operation, builds a data management platform, centrally stores, manages, and analyzes flaw detection data, provides support for operations, and effectively shortens the flaw detection time.

[0042] To effectively solve the problems of data fusion and data accuracy: Break through key technologies such as multi-modal data fusion, real-time noise filtering, and micro-defect classification, significantly different from the traditional manual experience interpretation mode, and achieve precise identification and dynamic optimization of defects under complex working conditions. The data of multiple sensors (lidar, ultrasonic, vision) are mutually verified to reduce the probability of false detection / missed detection. Innovatively utilize the Duffing chaotic oscillator technology to increase the signal-to-noise ratio by 40%, and use the deep learning framework to dynamically allocate the weights of multiple sensors to promote the leap of detection technology from "manual experience" to "data intelligence".

[0043] Advantages: 1. Advanced technological breakthrough - As a core means of non-destructive testing, traditional ultrasonic flaw detection is limited by single-frequency detection and manual experience-based interpretation, making it difficult to balance detection depth and accuracy. This device applies the Duffing chaotic oscillator subsystem and ultrasonic fusion technology, combined with the intelligent defect classification algorithm of machine deep learning, to overcome the problems of surface roughness scattering and deep crack signal attenuation of the 18-meter-long rail. It achieves a defect detection rate of ≥98%, with an accuracy improvement of over 50% compared to traditional methods. At the same time, the detection depth breaks through 200 mm, providing a full-dimensional damage diagnosis ability for scenarios such as high-speed railways and heavy-haul railways.

[0044] 2. Novel technical means - Traditional ultrasonic flaw detection relies on manual hand-held probes for point-by-point scanning, with low efficiency and high susceptibility to environmental factors. This device integrates multiple advanced technologies such as ultrasonic flaw detection, visual recognition flaw detection, gauge measurement, and intelligent monitoring, achieving all-round and multi-angle flaw detection of the track, greatly improving detection efficiency and accuracy. Through the Duffing chaotic oscillator subsystem, the detection speed and accuracy are significantly enhanced, and processes such as data acquisition and intelligent analysis are completed synchronously. In addition, a multi-source data fusion model of ultrasonic and machine vision is introduced to cross-verify ultrasonic signals and high-frequency image data, reducing the false judgment rate to less than 1% and breaking through the sensitivity limitations of single technologies in complex environments, providing a high-precision, high-efficiency, and high-life-cycle management solution for railway operation and maintenance.

[0045] Application effects: (1) Direct cost savings Labor cost reduction: One device reduces the annual labor demand by 5 people (from 6 to 1), saving 400,000 yuan per year; Maintenance cost optimization: Precise flaw detection extends the replacement cycle of the 18-meter-long rail to 18 months (from the original 12 months), saving 2 million yuan in annual material costs per 100 kilometers; Accident loss reduction: The hidden danger identification rate is increased to 98%, and the accident repair cost is reduced by 80% (saving an average of 500,000 yuan per line per year).

[0046] (2) Indirect benefit improvement Transport capacity release: The detection efficiency during the maintenance window is increased by 75%, and the annual number of additional train trips on high-speed railway lines is 500, with an additional income of over 100 million yuan; Data value addition: The health data of the 18-meter-long rail is connected to the railway operation and maintenance cloud platform, providing risk assessment services for the insurance and financial industries (market potential of 1 billion yuan per year).

[0047] Example 2: Such as Figure 1 - Figure 6As shown in the figure, this embodiment proposes an intelligent rail flaw detection device for implementing an intelligent rail flaw detection method in Embodiment 1. The intelligent rail flaw detection device includes a flaw detection trolley, which includes a chassis and a controller. The chassis is composed of a traveling cross beam 1 and two groups of traveling longitudinal beams 2 connected to the ends of the traveling cross beam 1; a vision system 3 and a laser rangefinder 19 are connected to the traveling cross beam 1, and an ultrasonic flaw detector 4 is provided on the traveling longitudinal beam 2. The vision system 3 is installed at the top of the adjusting rod, and the bottom of the adjusting rod is rotatably connected to the traveling cross beam 1, and the inclination angle can be adjusted and locked in position after adjustment.

[0048] A liquid storage tank 5 is detachably connected above the traveling longitudinal beam 2, and the coupling agent inside the liquid storage tank 5 is coated above the rail 18 through a spraying assembly 6.

[0049] A traveling system is arranged on the chassis and is used to drive the flaw detection trolley to move. The controller is respectively connected to electrical components such as the vision system 3, the ultrasonic flaw detector 4, the laser rangefinder 19, and the traveling system for control connection to perform real-time control.

[0050] The controller can transmit the flaw detection data to the observation device in real time through a wireless network, such as a mobile phone, a tablet or a computer, so that the staff can grasp the detection situation at any time. In addition, a power supply is also equipped on the traveling cross beam 1 to supply power to each power unit of the flaw detection trolley to ensure the stable operation of the device.

[0051] The whole device also includes two groups of dialing and discharging assemblies 7, both of which are arranged at the bottom of the traveling longitudinal beam 2 and act alternately to wipe off the liquid above the rail 18. The alternating dialing and discharging assembly 7 includes a driving motor 71 arranged inside the traveling longitudinal beam 2. The output shaft at the bottom of the driving motor 71 is connected to a swing arm 72, and a dial plate 73 is hinged to the bottom of the swing arm 72; it also includes an auxiliary rod 74 hinged to the bottom of the traveling longitudinal beam 2, and the other end of the auxiliary rod 74 is rotatably connected to the dial plate 73; when the driving motor 71 drives the swing arm 72 to rotate, it can drive the dial plate 73 to reciprocate and move outward. When the driving motor drives the swing arm 72 to rotate, it drives the dial plate 73 to move accordingly. By forming a constraint on one end of the dial plate 73 through the auxiliary rod 74, the other end of the dial plate 73 can move along the track line 75, so as to realize reciprocating motion and dial out the liquid on the upper surface of the rail 18.

[0052] In this embodiment, a support fixture 13 is fixedly connected above the traveling longitudinal beam 2, and a liquid storage tank 5 is placed above the support fixture 13. The liquid storage tank 5 is communicated with the spraying assembly 6 through a liquid discharge channel 14 at the bottom of the support fixture 13. This detachable method facilitates the connection between the liquid storage tank 5 and the support fixture 13, enabling the flaw detection liquid or coupling agent inside the liquid storage tank 5 to be coated above the rail 18 through the spraying assembly 6. The support fixture 13 has a rectangular structure, and grooves 15 are formed on four side walls of the support fixture 13. A limiting block 16 is hinged in the groove 15. The limiting block 16 has an L-shaped structure, and a clamping block 17 is fixedly connected to the top of the limiting block 16. A clamping groove 51 that can be engaged with the clamping block 17 is formed on the outer wall of the liquid storage tank 5. After the liquid storage tank 5 is placed in the support fixture 13, the bottom of the limiting block 16 is pressed, so that the clamping block 17 is engaged with the clamping groove 51, positioning the liquid storage tank 5 above the support fixture 13.

[0053] In some embodiments, the spraying assembly 6 includes an output pump 61 disposed inside the traveling longitudinal beam 2. The liquid discharge channel 14 is connected to the input end of the output pump 61. The spraying assembly 6 further includes a spraying box 62. A plurality of groups of nozzles 63 are fixed to the bottom of the spraying box 62. The output end of the output pump 61 is communicated with the nozzles 63 through the spraying box 62. During the coating stage, the liquid storage tank 5 is communicated with the liquid discharge channel 14, and the liquid in the liquid discharge channel 14 is conveyed to the inside of the spraying box 62 through the output pump 61, and the coupling agent is coated on the upper surface of the rail 18 by the nozzles 63 inside the spraying box 62.

[0054] In some embodiments, the traveling cross beam 1 includes a main beam body 101 and a connecting beam 102 located at the end of the main beam body 101. The main beam body 101 and the connecting beam 102 are detachably connected through a connecting component 103. This connecting piece adopts a conventional connection structure and will not be elaborated here. The connecting beam 102 is vertically arranged with the traveling longitudinal beam 2. After installation, the bottom of the flaw detection trolley as a whole has a T-shaped structure. The detachable connection between the main beam body 101 and the connecting beam 102 through the connecting piece enables the flaw detection trolley to be disassembled into two parts, which is convenient for storage or transportation, can reduce the space occupation, and can be quickly installed during use.

[0055] In some embodiments, traveling wheels 9 are respectively connected to the ends of the main beam body 101 and both ends of the traveling longitudinal beam 2. The traveling wheels 9 are arranged in a triangular pattern and are respectively located above two groups of rails 18, and are used to drive the traveling cross beam 1 and the traveling longitudinal beam 2 above them to move. Auxiliary positioning wheels 20 in a drum shape are respectively arranged inside the traveling wheels 9. The auxiliary positioning wheels 20 are movably arranged at the bottom of the traveling cross beam 1 or the traveling longitudinal beam 2 through springs and other components, and are convenient for installing and fitting against the side of the rail 18, avoiding phenomena such as derailment when the flaw detection trolley is traveling.

[0056] Technical effect: The coupling agent on the surface of the rail 18 is deflected to one side of the rail 18 by the deflector and rowing component 7. The two sets of deflector and rowing components 7 form a cycle of "wiping - resetting - wiping", achieving non-stop alternating operation, which can greatly improve the cleaning of the surface of the rail 18 and the uniform coating of the coupling agent, create better coupling conditions, and reduce the incidence of wave loss and missed detection problems.

[0057] Embodiment 3: On the basis of Embodiment 2, the whole machine further includes a recovery box 8. The recovery box 8 is fixed on both sides of the traveling longitudinal beam 2 and is used to collect the liquid deflected out of the rail 18 by the deflector plate 73. The movement track of the deflector and rowing component 7 is to move to both sides, which can deflect the liquid on the upper surface of the rail 18 into the recovery boxes 8 on both sides, realizing precise diversion, recovering the coupling agent on the upper surface of the rail 18, avoiding its residue on the rail 18, and also avoiding affecting the surrounding environment.

[0058] In some embodiments, the recovery box 8 is arranged in an inclined shape. A recovery pump 10 is fixedly connected to the lower edge of the recovery box 8. The recovery pump 10 transports the liquid in the recovery box 8 to the inside of the recovery tank 12 above the traveling longitudinal beam 2 through a recovery pipeline 11. The inclined design enables the liquid in the recovery box 8 to converge towards the lower edge under the action of gravity. Cooperating with the recovery pump 10 can accurately and efficiently extract the deflected liquid, reduce the residue, and transport it to the recovery tank 12 above the traveling longitudinal beam 2 through the recovery pipeline 11. After the detection process is completed, the waste liquid inside is then processed.

[0059] The following introduces the usage method of this device: Device assembly and preparation; Connect the traveling cross beam 1 of the flaw detection trolley to the connecting cross beam through the connecting component 103. Install the vision system 3 above the traveling cross beam 1, and install the ultrasonic flaw detector 4 at the bottom of the traveling longitudinal beam 2, and make the ultrasonic flaw detector 4 located between the liquid storage tank 5 and the deflector and rowing component 7; fix the liquid storage tank 5 filled with the coupling agent on the support fixture 13; Install one end of the traveling cross beam 1 in the flaw detection trolley above one side of the rail 18, install the traveling longitudinal beam 2 above the other side of the rail 18, and perform auxiliary positioning through the auxiliary positioning wheel 20 to ensure its traveling position.

[0060] Track detection; Start the spraying component 6, coat the coupling agent in the liquid storage tank 5 on the rail 18 through the spraying component 6, and smear the coupling agent on the surface of the rail 18 evenly through the deflector and rowing component 7; perform ultrasonic detection on the inside of the rail 18 through the ultrasonic component on the traveling longitudinal beam 2. At the same time, start the CMOS camera to continuously capture the surface image of the rail 18 at a high frame rate.

[0061] In summary, this device has the following technical advantages: Efficient hierarchical detection: Combining rapid screening with fine detection, taking into account both efficiency and accuracy.

[0062] Intelligent fusion algorithm: Dynamic weights and federated learning enhance the robustness of the system.

[0063] Full life cycle management: From positioning to defect tracking, supporting the optimization of track maintenance decisions.

[0064] Specifically, this intelligent device can achieve accurate detection and full-scenario coverage, reducing the workload of manual repeated detection and troubleshooting, lowering labor costs, improving environmental adaptability, promptly discovering and solving potential problems, and avoiding more serious damage and higher maintenance costs caused by the expansion of faults. The high integration, high precision, and high intelligence characteristics of this device have promoted the development of railway detection technology towards intelligence and precision, prompting related enterprises to continuously research and develop more advanced detection equipment and technologies. At the same time, it also provides data support for railway big data and intelligent operation and maintenance systems, promoting the development of railway technology.

[0065] The above are only preferred embodiments of the present invention, and thus do not limit the implementation manners and protection scope of the present invention. For those skilled in the art, it should be realized that all equivalent replacements and obvious changes made by using the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent rail flaw detection method, characterized in that, At least include a fast rough flaw detection mode, which includes machine vision detection and laser gauge measurement, where: The method of machine vision detection is as follows: Image acquisition: Capture the surface image of the rail through the vision system on the flaw detection trolley; Image preprocessing: Obtain the preprocessed rail image through noise reduction, enhancement, and segmentation; Defect analysis and classification: Input the preprocessed rail image into the YOLOV10 model. After in-depth analysis and processing, determine the surface defect type and obtain the rail surface defect detection result; The method of laser gauge measurement is as follows: Gauge data acquisition: Synchronously acquire gauge data through the laser rangefinder on the flaw detection trolley and mark the gauge overlimit area in real time; Data processing: Based on the image data after image preprocessing in machine vision detection and the gauge data in laser gauge measurement, fuse them through the edge computing module to generate a heat map of suspicious areas, and perform defect result display and defect annotation.

2. The intelligent rail flaw detection method according to claim 1, wherein It also includes a low-speed fine flaw detection mode, and the specific method is as follows: Ultrasonic data acquisition and preprocessing: Detect the defects inside the track through the ultrasonic flaw detector on the flaw detection trolley, acquire ultrasonic data and perform preprocessing to obtain the deep track defect detection result; Data processing: Input the image data after image preprocessing in machine vision detection and the ultrasonic data into the federated learning framework to dynamically optimize the defect classification model for processing, determine the defect type and classify it, generate the final track defect detection result and visualize the output.

3. The intelligent rail flaw detection method according to claim 2, characterized in that, In the ultrasonic data acquisition step, the Duffing chaotic oscillator subsystem is used to process the ultrasonic data, and the specific method is as follows: Signal acquisition: Use the ultrasonic probe of the ultrasonic flaw detector to obtain the echo signal; Chaotic system drive: Input the obtained echo signal into the Duffing system to reduce the noise interference of ultrasonic detection; Signal amplification and detection: The Duffing system detects the echo signal after noise reduction processing, and at the same time extracts and amplifies the features of the detected weak signal to obtain the processed echo signal data; Data analysis and processing: Analyze the processed echo signal data to determine the defect information.

4. The intelligent rail flaw detection method according to claim 1, characterized in that, The vision system uses a CMOS vision system.

5. An intelligent track flaw detection method according to claim 1, characterized in that, Defect types include but are not limited to cracks, wear, corrosion, and weld defects.

6. An intelligent track flaw detection device, characterized in that, An intelligent rail flaw detection method for implementing any one of claims 1-5, the intelligent rail flaw detection device includes a flaw detection trolley, including a chassis, the chassis is composed of a traveling cross beam and two groups of traveling longitudinal beams connected to the ends of the traveling cross beam; the traveling cross beam is connected with a vision system and a laser rangefinder, and an ultrasonic flaw detector is arranged on the traveling longitudinal beam; A liquid storage tank, detachably connected above the traveling longitudinal beam, and the coupling agent inside the liquid storage tank is coated above the rail through a spraying component; A traveling system, arranged on the chassis, for driving the flaw detection trolley to move.

7. An intelligent track flaw detection device according to claim 6, characterized in that, It also includes two groups of dialing and arranging components, both of which are arranged at the bottom of the traveling longitudinal beam and alternately act to wipe the liquid above the rail. The alternating dialing and arranging component includes a driving motor arranged inside the traveling longitudinal beam. The output shaft at the bottom of the driving motor is connected with a swing arm, and a dialing plate is hinged at the bottom of the swing arm. It also includes an auxiliary rod hinged at the bottom of the traveling longitudinal beam, and the other end of the auxiliary rod is rotatably connected with a dialing plate. When the driving motor drives the swing arm to rotate, it can drive the dialing plate to reciprocate and move outward.

8. An intelligent rail flaw detection device according to claim 6, characterized in that, A supporting fixture is fixedly connected above the traveling longitudinal beam, and the liquid storage tank is placed above the supporting fixture. The liquid storage tank is communicated with the spraying component through a liquid discharge channel at the bottom of the supporting fixture.

9. An integrated track defect detection device according to claim 8, characterized in that, The supporting fixture has a rectangular structure. Grooves are formed on the four side walls of the supporting fixture, and limiting blocks are hinged in the grooves. The limiting blocks have an L-shaped structure, and a clamping block is fixedly connected to the top of the limiting blocks. A clamping groove that can be clamped with the clamping block is formed on the outer wall of the liquid storage tank. After the liquid storage tank is placed in the supporting fixture, it presses the bottom of the limiting block, so that the clamping block is clamped with the clamping groove, and the liquid storage tank is positioned above the supporting fixture.

Images