1. The utility model provides a two formula sleeper appearance quality detecting system based on machine vision which characterized in that: the system comprises upper computer software in interactive communication with a PLC main program, a multi-axis servo linear guide rail, a sleeper transportation system and an installation frame; the upper computer software comprises a three-dimensional data acquisition control module, a three-dimensional data analysis processing module, an image data acquisition control module and an image data analysis processing module; the scanner and the tracker are controlled by the three-dimensional data acquisition control module and the three-dimensional data analysis processing module, and the industrial camera and the inkjet printer are controlled by the image data acquisition control module and the image data analysis processing module;

the multi-axis servo linear guide rail comprises an X-axis servo linear guide rail, a Y-axis servo linear guide rail and a tracking servo linear guide rail; a mechanical arm is fixedly hung on a base of the X-axis servo linear guide rail in an inverted mode, a scanner is installed on the mechanical arm, and a tracker is installed on the tracking servo linear guide rail; the PLC main program controls the multi-axis servo linear guide rail and the mechanical arm to operate and is in interactive communication with the sleeper transportation system.

2. The system of claim 1, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the upper computer software is connected with a PLC main program, a tracker and a scanner through a communication transmission protocol TCP/IP; the upper computer software controls the start and stop of scanning, automatically acquires scanning data and automatically analyzes and processes the acquired scanning data; the upper computer software controls the industrial camera to photograph the sleeper, carries out crack detection on the sleeper picture and acquires the model code information of the sleeper; and the upper computer software controls the code spraying machine to code the sleepers which have finished the detection task.

3. The system of claim 1, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the sleeper transport system comprises an AVG trolley, a sleeper transport drag chain line, a blocking device, a jacking device and a sensor; the AVG trolley is used for placing the sleeper on a sleeper transportation drag chain line, and after the sleeper transportation drag chain line transports the sleeper to the blocking device, the jacking device jacks the sleeper to the detection position.

4. The system of claim 3, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the sleeper transportation system ensures that the repeated positioning precision of each group of sleepers is within 1cm, and meets the requirement of three-dimensional scanning.

5. The system of claim 3, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the PLC main program is connected and communicated with the mechanical arm, the industrial camera and the ink-jet printer; the PLC main program receives a feedback signal through the sensor, controls an industrial camera to shoot and collect images, and controls an ink-jet printer to execute code-jet operation; the PLC main program establishes a motion control program to control the multi-axis servo linear guide rail and the mechanical arm to move according to a set path; and the mechanical arm is provided with a controller carrying a fine scanning program.

6. The system of claim 1, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the mounting frame comprises a truss type I-steel main frame, a positioning mark point support, a tracker support capable of adjusting the angle and the direction, a camera mounting support and an ink-jet printer support; an industrial camera, a light source and a photographing triggering sensor are arranged on the camera mounting bracket; and the ink-jet printer bracket is provided with an ink-jet printer, a code meter and a code-jet triggering sensor.

7. The system of claim 6, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the vertical precision of the installation of the truss type I-steel main frame is better than 1nm, and the horizontal precision is better than 3 nm; the vertical precision and the horizontal precision of the multi-axis servo linear guide rail are both superior to 1nm, and the included angle precision of an X axis and a Y axis is superior to 0.05 degrees.

8. The system of claim 1, wherein the system is configured to detect the appearance quality of the dual block tie based on machine vision, and further configured to: the industrial camera is 2500 ten thousand pixels high resolution and is used for collecting sleeper images and identifying cracks, width and length of a sleeper groove; the system utilizes deep learning based OCR technology to identify tie mold numbers.

9. A double-block sleeper appearance quality detection method based on machine vision is characterized by comprising the following steps:

s1, starting hardware and software of the detection system, carrying out system initialization and equipment communication connection, and enabling the system to be in a standby state;

s2, conveying and feeding sleepers, placing a group of sleepers on a sleeper conveying drag chain line by an AGV trolley, lifting a blocking device after a sensor detects a signal, triggering the sensor signal again after the sleepers are conveyed to the blocking device and are completely closed, stopping the operation of the sleeper conveying drag chain line, and falling the blocking device; the method comprises the following steps that a sleeper is lifted by a lifting device, and after the sleeper is lifted to the proper position, a towline control system sends a sleeper proper position signal to a PLC main program of a detection system;

s3, three-dimensional scanning detection is carried out, a PLC main program sends a sleeper in-place signal to upper computer software, and the upper computer software starts a scanner; after the scanner is started successfully, the upper computer software sends a scanning starting signal to the PLC main program, and the PLC main program executes a motion program after receiving the signal, controls the multi-axis servo linear guide rail and the mechanical arm, and starts three-dimensional scanning according to a set path; after the motion program is executed, the PLC main program sends a scanning completion signal to a towline control system to inform a sleeper jacking device on a sleeper transportation towline to descend; meanwhile, a scanning completion signal is sent to the upper computer, the upper computer software stops the scanner, three-dimensional point cloud data of the sleeper are automatically obtained for analysis and processing, and meanwhile, a three-dimensional detection result is stored; the PLC main program controls the multi-axis servo linear guide rail and the mechanical arm to return to a mechanical original point;

s4, carrying out appearance detection; after the sleeper jacking device falls down, the sleeper is continuously transported forwards by the sleeper transporting drag chain line, and the sensor is triggered when the sleeper enters the appearance detection area, so that the cleaning device is automatically started; entering the visual field range of the industrial camera after about 10 seconds, and sending a photographing signal to the upper computer software by the PLC main program; after receiving the photographing signal, the upper computer software starts a camera to photograph, acquires image data to analyze and process, automatically identifies sleeper cracks and sleeper numbers, controls an ink-jet printer to automatically jet ink and stores crack identification results;

the S5 tie transport towline continues to run forward transporting ties to the next operating station.

10. The machine vision-based method for detecting the appearance quality of the double-block sleeper, as claimed in claim 9, is characterized in that: the steps S1 to S5 are a complete process for inspecting one group of sleepers, and the next group of sleepers enters the inspection process from S2.

Background

The traditional double-block type sleeper detection technology is mainly completed by adopting a steel ruler, a depth gauge, a vernier caliper, a footing distance measuring instrument, a hole inclinometer, a filler gauge, a relative distortion detector and the like, about 5-10 minutes is needed for detecting a group of (4) double-block type sleepers, the detection efficiency is low, the labor is consumed, the detection requirement cannot be met, and the informatization of the railway which is vigorously propelled by a railway head office cannot be met. According to the requirement of detection specifications, the processing errors of part of detection items are less than 0.5mm, and the detection precision of the traditional method cannot completely meet the error requirement of size detection, so that the research on a new detection technology has important engineering application value. With the continuous development of three-dimensional laser scanning technology and deep learning technology in recent years, three-dimensional laser scanners and high-resolution industrial cameras with different accuracies and different performances are on the market more and more, and as three-dimensional laser scanning has the advantages of non-contact, high accuracy, high efficiency, complete data and the like, the high-accuracy three-dimensional laser scanners are more and more applied to the field of industrial measurement, for example: the method comprises the following steps of measuring the sizes of automobile, aviation and mechanical parts, measuring cultural relics and antiques and sculpture artworks, designing forward and reverse industries, displaying three-dimensional products and the like. Deep learning is applied to a wide range of fields, such as face recognition, automatic driving, safety precaution, pedestrian detection, classification detection of remote sensing images, and the like.

Considering that the directions of cracks of the double-block sleeper are various and the double-block sleeper is difficult to detect by ordinary means, in order to further improve the efficiency and realize a series of functions of full-automatic size measurement, crack detection and data calculation, a high-precision appearance size detection system suitable for the high-speed rail double-block sleeper needs to be designed.

Disclosure of Invention

The invention provides a double-block sleeper appearance quality detection system and method based on machine vision for solving the technical problems in the known technology, a deep learning technology is adopted to detect cracks of an image acquired by an industrial camera, information such as a mold number code is extracted, and after the system is installed and debugged, streamlined and automatic detection of a sleeper can be realized.

The invention comprises the following technical scheme: a double-block sleeper appearance quality detection system based on machine vision comprises upper computer software interactively communicating with a PLC main program, a multi-axis servo linear guide rail, a sleeper transportation system and an installation frame;

the upper computer software comprises a three-dimensional data acquisition control module, a three-dimensional data analysis processing module, an image data acquisition control module and an image data analysis processing module; the scanner and the tracker are controlled by the three-dimensional data acquisition control module and the three-dimensional data analysis processing module, and the industrial camera and the inkjet printer are controlled by the image data acquisition control module and the image data analysis processing module;

the multi-axis servo linear guide rail comprises an X-axis servo linear guide rail, a Y-axis servo linear guide rail and a tracking servo linear guide rail; a mechanical arm is fixedly hung on a base of the X-axis servo linear guide rail in an inverted mode, a scanner is installed on the mechanical arm, and a tracker is installed on the tracking servo linear guide rail; the PLC main program controls the multi-axis servo linear guide rail and the mechanical arm to operate to drive the scanner and the tracker to acquire three-dimensional data, and the PLC main program is in interactive communication with the sleeper transportation system, so that the system is compatible with the production rhythm of a production line.

Further, the upper computer software is connected with a PLC main program, a tracker and a scanner through a communication transmission protocol TCP/IP; the upper computer software controls the start and stop of scanning, automatically acquires scanning data and automatically analyzes and processes the acquired scanning data; the upper computer software controls the industrial camera to photograph the sleeper, carries out crack detection on the sleeper picture and acquires the model code information of the sleeper; and the upper computer software controls the code spraying machine to code the sleepers which have finished the detection task.

Further, the sleeper transport system comprises an AVG trolley, a sleeper transport drag chain line, a blocking device, a jacking device and a sensor; the sensor is arranged below the sleeper transport drag chain line. The AVG trolley is used for placing the sleeper on a sleeper transportation drag chain line, and after the sleeper transportation drag chain line transports the sleeper to the blocking device, the jacking device jacks the sleeper to the detection position. The blocking device and the jacking device are arranged on two sides of a sleeper transportation drag chain line, and the blocking device is located at the front end of the jacking device and is located at a position far away from a material loading position.

The stop device comprises a stop block, a stop block lifting support and a lifting cylinder, wherein the stop block is fixedly arranged on the top surface of the stop block lifting support, the lifting cylinder is connected with a movable rod of the stop block lifting support through a connecting rod, and the lifting cylinder enables the stop block to lift through pushing the movable rod to rotate. The bottom of the stop block lifting support is fixedly arranged on the top surface of the mounting frame, and the output end of the lifting air cylinder penetrates through the middle part of the mounting frame to be connected with the connecting rod.

The jacking device comprises a jacking support and a horizontal cylinder, the horizontal cylinder is connected with the bottom edge of the jacking support, and the movable rod of the jacking support rotates by pushing and pulling the bottom edge, so that the top surface of the jacking support descends or rises. The movable rod of the jacking bracket is arranged on the side surface of the sleeper transportation drag chain line through a rotating shaft. The sensor is installed below a sleeper transportation drag chain line on the inner side of the jacking device and used for detecting the sleeper.

The system is installed on a sleeper production line in an integrated mode, and the operation of the sleeper transportation drag chain line is controlled by a drag chain line control system on the original production line. Trigger the sensor when the sleeper is placed the sleeper transportation tow chain from the AGV dolly and make blocking device rise, the sleeper transports to blocking device along with sleeper transportation tow chain, puts together a set of (4) sleepers, and the sleeper is put together the back, triggers the sensor once more and makes blocking device descend, and jacking device jacks up the sleeper, makes the sleeper reach the detection position.

Furthermore, the sleeper transportation system ensures that the repeated positioning precision of each group of sleepers is within 1cm, meets the requirement of three-dimensional scanning, and avoids complex coordinate system calibration and sleeper positioning.

Further, the PLC main program is connected with the mechanical arm, the industrial camera and the ink-jet printer for communication; the PLC main program receives the feedback signal through the sensor, controls the industrial camera to shoot and collect images, and controls the code spraying machine to execute code spraying operation. The system adopts an automatic code spraying machine to finish automatic code spraying and writing. The PLC main program establishes a motion control program to control the multi-axis servo linear guide rail and the mechanical arm to move according to a set path. And the mechanical arm is provided with a controller carrying a fine scanning program.

Further, the mounting frame comprises a truss type I-steel main frame, a positioning mark point support, a tracker support capable of adjusting the angle and the direction, a camera mounting support and an ink-jet printer support; an industrial camera, a light source and a photographing triggering sensor are arranged on the camera mounting bracket; and the ink-jet printer bracket is provided with an ink-jet printer, a code meter and a code-jet triggering sensor.

The plane of the truss type I-steel main frame is rectangular, sleeper transportation drag chain lines are arranged on the ground below the rectangle, and stand columns are arranged at four corner points of the rectangle; two parallel sides of the rectangle are provided with Y-axis servo linear guide rails, and X-axis servo linear guide rails are vertically arranged on the two Y-axis servo linear guide rails; one is installed on the ground below the servo linear guide rail of Y axle and trails servo linear guide rail, another is installed on the ground below the servo linear guide rail of Y axle and fixes a position mark point support, install the location mark point on the location mark point support, tracker and location mark point all are on a parallel with sleeper transportation drag chain line, carry out accurate measurement and location to the sleeper through the two. The camera mounting bracket is parallel to the X-axis servo linear guide rail and fixedly mounted on the side face of the upright post, and the camera mounting bracket is positioned on one side of the sleeper away from the range of the truss type I-steel main frame.

Furthermore, the vertical precision of the installation of the truss type I-steel main frame is better than 1nm, and the horizontal precision is better than 3 nm; the vertical precision and the horizontal precision of the multi-axis servo linear guide rail are both superior to 1nm, and the included angle precision of an X axis and a Y axis is superior to 0.05 degrees.

Furthermore, the tracker and the scanner are three-dimensional laser scanning equipment, and are integrated and developed based on SDK provided by manufacturers, so that upper computer software can automatically control the equipment to acquire data. In the running process of the system, after the multi-axis servo linear guide rail moves in place according to the debugging position, the PLC main program controls the mechanical arm to do fine scanning movement according to the debugging path, and the required data is obtained through scanning.

Furthermore, the industrial camera has the high resolution of 2500 ten thousand pixels, is used for collecting sleeper images and identifying cracks, width and length of a sleeper groove, and has the characteristics of high identification precision and good stability. The system identifies the sleeper mold number by using an OCR technology based on deep learning, can adapt to complex illumination conditions, and has the identification accuracy rate of 99%.

A double-block sleeper appearance quality detection method based on machine vision comprises the following steps:

s1, starting hardware and software of the detection system, carrying out system initialization and equipment communication connection, and enabling the system to be in a standby state;

s2, conveying and feeding sleepers, placing a group of sleepers on a sleeper conveying drag chain line by an AGV trolley, lifting a blocking device after a sensor detects a signal, triggering the sensor signal again after the sleepers are conveyed to the blocking device and are completely closed, stopping the operation of the sleeper conveying drag chain line, and falling the blocking device; the method comprises the following steps that a sleeper is lifted by a lifting device, and after the sleeper is lifted to the proper position, a towline control system sends a sleeper proper position signal to a PLC main program of a detection system;

s3, three-dimensional scanning detection is carried out, a PLC main program sends a sleeper in-place signal to upper computer software, and the upper computer software starts a scanner; after the scanner is started successfully, the upper computer software sends a scanning starting signal to the PLC main program, and the PLC main program executes a motion program after receiving the signal, controls the multi-axis servo linear guide rail and the mechanical arm, and starts three-dimensional scanning according to a set path; after the motion program is executed, the PLC main program sends a scanning completion signal to a towline control system to inform a sleeper jacking device on a sleeper transportation towline to descend; meanwhile, a scanning completion signal is sent to the upper computer, the upper computer software stops the scanner, three-dimensional point cloud data of the sleeper are automatically obtained for analysis and processing, and meanwhile, a three-dimensional detection result is stored; the PLC main program controls the multi-axis servo linear guide rail and the mechanical arm to return to a mechanical original point;

s4, carrying out appearance detection; after the sleeper jacking device falls down, the sleeper is continuously transported forwards by the sleeper transporting drag chain line, and the sensor is triggered when the sleeper enters the appearance detection area, so that the cleaning device is automatically started; entering the visual field range of the industrial camera after about 10 seconds, and sending a photographing signal to the upper computer software by the PLC main program; after receiving the photographing signal, the upper computer software starts a camera to photograph, acquires image data to analyze and process, automatically identifies sleeper cracks and sleeper numbers, controls an ink-jet printer to automatically jet ink and stores crack identification results;

the S5 tie transport towline continues to run forward transporting ties to the next operating station.

Further, the steps from S1 to S5 are a complete process for inspecting one group of sleepers, and the next group of sleepers enters the inspection process from S2, and the whole process does not need any manual intervention.

The invention has the advantages and positive effects that:

1. the upper computer software adopts three-dimensional laser scanning and three-dimensional point cloud processing technology, and automatically detects the outline dimension of the sleeper under the assistance of an AVG trolley and a sleeper drag chain line; and automatically acquiring the three-dimensional point cloud, automatically analyzing and processing the three-dimensional point cloud and automatically outputting an appearance detection report.

2. The invention adopts a high-resolution industrial camera and an image depth learning technology to automatically obtain a high-resolution image, and can automatically identify the sleeper code and automatically detect the surface cracks of the sleeper.

3. The upper computer software performs unified control on all hardware equipment through the PLC main program, can complete detection of multiple indexes of the sleeper in a highly automatic and collaborative manner, and has the advantages of high integration and automation; in addition, the system has the advantages of high size detection precision of 0.1mm and crack detection precision of 0.2mm, and high precision.

4. The invention utilizes the sleeper transport line, the blocking device and the jacking device to realize accurate repeated positioning of each group of sleepers, the positioning precision is about 1-2cm, and the requirement of three-dimensional scanning is met; after the sleeper is in place, the detection system is started, high-precision three-dimensional point cloud of the sleeper is obtained through the tracker, the scanner and the positioning mark points, and size analysis is carried out based on the three-dimensional point cloud.

Drawings

Fig. 1 is a control flow chart of the present invention.

Fig. 2 is a schematic front view of the present invention.

FIG. 3 is a schematic sectional view A-A of the present invention.

Fig. 4 is a side view schematic of the present invention.

Fig. 5 is a schematic structural view of the blocking device and the jacking device.

Fig. 6 is a schematic view of the structure of the blocking device when raised.

Fig. 7 is a schematic perspective view of the present invention.

In the figure, 1-control cabinet; 2-X axis servo linear guide rail; 3-Y axis servo linear guide rail; 4-tracking the servo linear guide rail; 5, a mechanical arm; 51-a controller; 6-truss type I-steel main frame; 61-upright post; 7-tracker mount; 8-scanner transfer flange; 9-a tracker; 10-a scanner; 11-locating the landmark points; 12-positioning the landmark point holder; 13-an industrial camera; 14-an ink jet printer; 15-sleeper transport towline; 16-a sleeper; 17-a stop block; 18-a stopper lifting bracket; 19-a lifting cylinder; 20-jacking a support; 21-horizontal cylinder; 22-sensor.

Detailed Description

To further clarify the disclosure of the present invention, its features and advantages, reference is made to the following examples taken in conjunction with the accompanying drawings.

Example (b): referring to fig. 1-7, a machine vision-based dual-block sleeper appearance quality detection system includes upper computer software interactively communicating with a PLC main program, a multi-axis servo linear guide rail, a sleeper transport system, and an installation frame; host computer software and PLC main program are installed in switch board 1, and switch board 1 is inside still to be equipped with industrial computer, switch and power, and the touch-sensitive screen operation is passed through on the surface of switch board 1.

As shown in fig. 1, the upper computer software includes a three-dimensional data acquisition control module, a three-dimensional data analysis processing module, an image data acquisition control module, and an image data analysis processing module; the scanner 10 and the tracker 9 are controlled by a three-dimensional data acquisition control module and a three-dimensional data analysis processing module, and the industrial camera 13 and the inkjet printer 14 are controlled by an image data acquisition control module and an image data analysis processing module; the upper computer software is connected with a PLC main program, a tracker 9 and a scanner 10 through a communication transmission protocol TCP/IP; the upper computer software controls the start and stop of scanning, automatically acquires scanning data and automatically analyzes and processes the acquired scanning data; the upper computer software controls the industrial camera 13 to photograph the sleeper, crack detection is carried out on the sleeper photograph, and model code information of the sleeper is obtained; the upper computer software controls the code spraying machine 14 to code the sleepers which have finished the detection task.

The PLC main program is connected with the mechanical arm 5, the industrial camera 13 and the ink-jet printer 14 for communication; the PLC main program receives a feedback signal through a sensor, controls the industrial camera 13 to shoot and collect images, and controls the code spraying machine 14 to execute code spraying operation. The system adopts an automatic code spraying machine to finish automatic code spraying and writing. The PLC main program establishes a motion control program to control the multi-axis servo linear guide rail and the mechanical arm 5 to move according to a set path. The robot arm 5 is provided with a controller 51 on which a fine scanning program is loaded.

As shown in fig. 2 to 7, the multi-axis servo linear guide includes an X-axis servo linear guide 2, a Y-axis servo linear guide 3, and a tracking servo linear guide 4; a mechanical arm 5 is fixedly hung upside down on a base of the X-axis servo linear guide rail 2, a scanner 10 is installed on the mechanical arm 5 through a scanner adapter flange 8, and a tracker 9 is installed on the tracking servo linear guide rail 4 through a tracker bracket 7; the PLC main program controls the multi-axis servo linear guide rail and the mechanical arm 5 to operate to drive the scanner 10 and the tracker 9 to acquire three-dimensional data, and the three-dimensional data are interactively communicated with the sleeper transportation system, so that the system is compatible with the production rhythm of a production line.

The sleeper transport system comprises an AVG trolley, a sleeper transport drag chain line 15, a blocking device, a jacking device and a sensor 22; the sensor 22 is arranged below the sleeper transport towline 15. The AVG trolley is used for placing the sleeper on the sleeper transportation drag chain line 15, and after the sleeper transportation drag chain line 15 transports the sleeper to the blocking device, the jacking device jacks the sleeper to the detection position. The blocking device and the jacking device are arranged on two sides of the sleeper transportation drag chain line 15, and the blocking device is located at the front end of the jacking device and is located at a position far away from a material loading position.

The stopping device comprises a stop block 17, a stop block lifting support 18 and a lifting cylinder 19, wherein the stop block 17 is fixedly arranged on the top surface of the stop block lifting support 18, the lifting cylinder 19 is connected with a movable rod of the stop block lifting support 18 through a connecting rod, and the lifting cylinder 19 enables the stop block 17 to lift through pushing the movable rod to rotate. The bottom of the stop block lifting support 18 is fixedly arranged on the top surface of the mounting frame, and the output end of the lifting cylinder 19 penetrates through the middle part of the mounting frame to be connected with the connecting rod.

The jacking device comprises a jacking bracket 20 and a horizontal cylinder 21, wherein the horizontal cylinder 21 is connected with the bottom edge of the jacking bracket 20, and the movable rod of the jacking bracket 20 rotates by pushing and pulling the bottom edge, so that the top surface of the jacking bracket 20 descends or rises. The movable rod of the jacking bracket 20 is installed at the side of the sleeper transport drag chain line 15 through a rotating shaft. The sensor 22 is mounted below the sleeper transport drag chain line 15 inside the jacking device for detecting the sleepers 16.

The system is installed on a sleeper production line in an integrated mode, and the operation of the sleeper transportation drag chain line 15 is controlled by a drag chain line control system on the original production line. When the sleeper is placed on the sleeper transportation drag chain line 15 from the AGV trolley, the sensor 22 is triggered to lift the blocking device, the sleeper is transported to the blocking device along with the sleeper transportation drag chain line 15, a group of (4) sleepers are gathered together, after the sleeper is gathered together, the sensor 22 is triggered again to lower the blocking device, and the jacking device jacks the sleeper so that the sleeper reaches the detection position. The sleeper transportation system ensures that the repeated positioning precision of each group of sleepers is within 1cm, meets the requirement of three-dimensional scanning, and avoids complex coordinate system calibration and sleeper positioning.

The mounting frame comprises a truss type I-steel main frame 6, a positioning mark point support 12, a tracker support 7 with adjustable angle and direction, a camera mounting support and an ink-jet printer support. The plane of the truss type I-steel main frame 6 is rectangular, sleeper transportation drag chain lines 15 are arranged on the ground below the rectangle, and stand columns 61 are arranged at four corner points of the rectangle; two parallel sides of the rectangle are provided with Y-axis servo linear guide rails 3, and X-axis servo linear guide rails 2 are vertically arranged on the two Y-axis servo linear guide rails 3; one of the two linear guide rails is provided with a tracking servo linear guide rail 4 on the ground below the Y-axis servo linear guide rail 3, the other one of the two linear guide rails is provided with a positioning mark point support 12 on the ground below the Y-axis servo linear guide rail 3, the positioning mark point support 12 is provided with a positioning mark point 11, and the tracker 9 and the positioning mark point 11 are both parallel to a sleeper transportation drag chain line 15, so that accurate measurement and positioning are carried out on the sleeper 16 through the two linear guide rails.

The camera mounting bracket is parallel to the X-axis servo linear guide rail 2 and is fixedly arranged on the side surface of the upright post 61, and the camera mounting bracket is positioned on one side of the sleeper 16 away from the range of the truss type I-steel main frame 6. The camera mounting bracket is provided with an industrial camera 13, a light source and a photographing trigger sensor; and the ink-jet printer support is provided with an ink-jet printer 14, a code meter and a code-jet trigger sensor. The vertical precision of the installation of the truss type I-steel main frame 6 is better than 1nm, and the horizontal precision is better than 3 nm; the vertical precision and the horizontal precision of the multi-axis servo linear guide rail are both superior to 1nm, and the included angle precision of an X axis and a Y axis is superior to 0.05 degrees.

The tracker 9 and the scanner 10 are three-dimensional laser scanning devices, and are integrated and developed based on SDK provided by manufacturers, so that upper computer software can automatically control the devices to acquire data. In the running process of the system, after the multi-axis servo linear guide rail moves in place according to the debugging position, the PLC main program controls the mechanical arm 5 to do fine scanning movement according to the debugging path, and the required data is obtained through scanning.

The industrial camera 13 has a high resolution of 2500 ten thousand pixels, is used for collecting sleeper images and identifying cracks, width and length of a sleeper groove, and has the characteristics of high identification precision and good stability. The system identifies the sleeper mold number by using an OCR technology based on deep learning, can adapt to complex illumination conditions, and has the identification accuracy rate of 99%.

The working process is as follows: s1, starting hardware and software of the detection system, carrying out system initialization and equipment communication connection, and enabling the system to be in a standby state;

s2 sleeper transportation feeding is carried out, an AGV trolley puts a group of sleepers on the sleeper transportation drag chain line 15, the sensor detects a signal and then the blocking device rises, the sleepers are transported to the blocking device and are completely closed, then the sensor signal is triggered again, the sleeper transportation drag chain line 15 stops running, and the blocking device falls down; the method comprises the following steps that a sleeper is lifted by a lifting device, and after the sleeper is lifted to the proper position, a towline control system sends a sleeper proper position signal to a PLC main program of a detection system;

s3, three-dimensional scanning detection is carried out, a sleeper in-place signal is sent to upper computer software by a PLC main program, and the upper computer software starts the scanner 10; after the scanner 10 is started successfully, the upper computer software sends a scanning starting signal to the PLC main program, and the PLC main program executes a motion program after receiving the signal, controls the multi-axis servo linear guide rail and the mechanical arm 5, and starts three-dimensional scanning according to a set path; after the motion program is executed, the PLC main program sends a scanning completion signal to the towline control system to inform the sleeper lifting device on the sleeper transportation towline 15 to descend; meanwhile, a scanning completion signal is sent to the upper computer, the upper computer software stops the scanner 10, three-dimensional point cloud data of the sleeper are automatically obtained for analysis and processing, and meanwhile, a three-dimensional detection result is stored; the PLC main program controls the multi-axis servo linear guide rail and the mechanical arm 5 to return to the original point of the machine;

s4, carrying out appearance detection; after the sleeper jacking device falls down, the sleeper is continuously transported forwards by the sleeper transporting drag chain line, and the sensor is triggered when the sleeper enters the appearance detection area, so that the cleaning device is automatically started; entering the visual field range of the industrial camera 13 after about 10 seconds, and sending a photographing signal to the upper computer software by the PLC main program; after receiving the photographing signal, the upper computer software starts a camera to photograph, acquires image data to analyze and process, automatically identifies sleeper cracks and sleeper numbers, controls the code spraying machine 14 to automatically spray codes and stores crack identification results;

the S5 tie transport towline continues to run forward transporting ties to the next operating station.

The steps from S1 to S5 are a complete process for detecting one group of sleepers, and the next group of sleepers enter the detection process from S2, and the whole process does not need any manual intervention.

While the preferred embodiments of the present invention have been illustrated and described, it will be appreciated by those skilled in the art that the foregoing embodiments are illustrative and not limiting, and that many changes may be made in the form and details of the embodiments of the invention without departing from the spirit and scope of the invention as defined in the appended claims. All falling within the scope of protection of the present invention.