A rail profile dynamic detection device and a profile data splicing method

Abstract

The application provides a steel rail profile dynamic detection device and a profile data splicing method, and belongs to the field of steel rail profile detection. The device comprises a left profile measurement unit, a right profile measurement unit, a main body walking unit, a connecting cross beam, an auxiliary walking unit, a manual push rod unit, a left side limiting unit, a right side limiting unit and a PC assembly. The manual push rod unit is used for collecting steel rail profile data in a dynamic manner, and a high-precision encoder wheel assembly is arranged. In the process of walking, the steel rail profile data is sampled once at a high sampling precision (one pulse is triggered every 0.5mm of walking), so that the problem that the traditional detection device cannot continuously sample is solved, and the working efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of rail profile detection, and particularly relates to a rail profile dynamic detection device and a profile data splicing method. BACKGROUND

[0002] Wheel-rail relationship is a key factor affecting the safe operation and comfort of railway vehicles, and rail profile is an important part of the wheel-rail relationship. Since the change of the rail profile is directly related to the safe operation of the railway track, the rail profile detection helps to master the service state of the rail and is an important means for railway operation and maintenance.

[0003] The traditional rail profile detection method is rail profile instrument detection. The operator measures the rail profile data with the rail profile instrument, and then compares the data with the design drawing or industry standard to judge the service state of the rail and provide data support for the rail health management system. This measurement method has the problems of high work intensity, inability to continuous measurement, low work efficiency, and great influence of subjective factors.

[0004] To solve the above problems, people have been seeking an effective method. SUMMARY

[0005] In view of the above shortcomings in the prior art, the rail profile dynamic detection device and the profile data splicing method provided by the present application solve the problem that the traditional detection device cannot continuously sample.

[0006] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: a rail profile dynamic detection device, comprising a left profile measurement unit, a right profile measurement unit, a main body walking unit, a connecting beam, an auxiliary walking unit, a manual push rod unit, a left side limiting unit, a right side limiting unit and a PC component;

[0007] The left profile measurement unit and the right profile measurement unit are both installed on the main body walking unit, the left side limiting unit, the right side limiting unit and the PC component are all installed on the connecting beam, the manual push rod unit is installed on the top of the connecting beam, one end of the connecting beam is connected with the auxiliary walking unit, and the other end of the connecting beam is connected with the main body walking unit and installed on the connecting beam.

[0008] The beneficial effects of the present application are: the present application uses left and right measuring units to ensure the integrity of the rail profile detection data; the limiting unit is used to limit the left and right directions of the equipment to prevent the left and right skewing of the equipment during the travel; the connecting cross beam ensures that the measuring unit does not skew during the detection of the rail; the manual pushing rod assembly is used to transmit force to the device main body walking unit to drive the walking unit to travel, thereby realizing the travel of the device along the rail direction; the present application dynamically collects rail profile data through the manual pushing rod unit, and is equipped with a high-precision encoder wheel assembly, which performs rail profile data sampling work at a high sampling accuracy (0.5mm per travel triggers a pulse) during the travel, effectively solving the problem that the traditional detection device cannot continuously sample, and improving the work efficiency.

[0009] The present application also provides a profile data splicing method, comprising the following steps:

[0010] S1, set up the rail profile dynamic detection device on the rail section, tightly attach the limiting wheels of the left and right limiting units to the inner sides of the two rails, and start the rail profile dynamic detection device;

[0011] S2, push the manual pushing rod to drive the dynamic detection device to travel along the rail in the forward direction, trigger the rail profile dynamic detection device to perform sampling work, and scan the rail profile data for processing; if the current rail section has not been fully collected, continue to push the manual pushing rod to drive the dynamic detection device to travel along the rail in the forward direction, and trigger the rail profile dynamic detection device to continue the sampling work;

[0012] S3, based on the processed rail profile data, perform profile registration processing to obtain an optimal translation rotation matrix;

[0013] S4, splice the rail profile data based on the optimal translation rotation matrix to obtain the spliced rail profile data, wherein the comparison result of the spliced rail profile data with the standard rail profile data in the database is used to judge the service state of the measured rail.

[0014] The beneficial effects of the present application are: the line laser sensors are symmetrically erected on both sides of the rail to cover the inner and outer sides of the rail. This identification method provides effective data support for profile registration and splicing, and the detection data is comprehensive and reliable. At the same time, the data processing method based on moving average and grey correlation analysis is used to complete the rail profile data splicing, which realizes high registration accuracy in the case of close initial position, and judges the registration effect by analyzing the correlation degree of the profile overlap range on both sides. The algorithm has fast calculation time, high work efficiency, good profile data registration effect, and high data reliability.

[0015] Further, the S2 comprises the following steps:

[0016] S201, pushing the artificial push rod to drive the dynamic detection device to move along the steel rail in the forward direction, and scanning the steel rail profile data by using the line laser sensor;

[0017] S202, removing the abnormal points of the scanned steel rail profile data by using the moving average method, and completing the processing of the steel rail profile data.

[0018] The beneficial effects of the above further scheme are: the artificial push rod drives the dynamic detection device to move along the steel rail in the forward direction, and the moving average method is used to remove the abnormal points in the data, so that the detection data is more reliable and effective.

[0019] Further, the S202 includes the following steps:

[0020] Based on the scanned steel rail profile data, the mean value of the previous k-1 points and the current point is calculated for each steel rail profile data point in the window, the moving average value is obtained, and the previous k-1 points of the steel rail profile data in the window are set to null;

[0021] Based on the moving average value, the standard deviation is calculated;

[0022] The mean value of all standard deviations is calculated , the standard deviation and the threshold value , if the standard deviation exceeds , it is determined as a jump point, and the jump point is removed, and the data of the steel rail profile data is completed.

[0023] The beneficial effects of the above further scheme are: the abnormal points are removed by judging the jump points, so that the detection data is more reliable and effective.

[0024] Further, the expression of the moving average value is as follows:

[0025] ;

[0026] Wherein, represents the moving average value, t represents the number of steel rail profile data points, represents the first i steel rail profile data point.

[0027] Further, the S3 includes the following steps:

[0028] S301, based on the position of the line laser sensor, the matrix relationship between the line laser sensor coordinate system and the space coordinate system is calculated, the relationship between the line laser sensor measurement image data and the space coordinate system is derived, and the line laser sensor is preliminarily translated and rotated;

[0029] S302. Initialize the translation and rotation matrices;

[0030] S303. Based on the processed rail profile data, retain the overlap range of the rail profile data on both sides in the rail profile data obtained in each translation and rotation, where the horizontal coordinates of the two rail profile data are the same at this time.

[0031] S304. Use the ordinate values ​​of the non-working side coincident rail profile data as a reference sequence. , recorded as And the ordinate of the rail profile data with overlapping working sides is used as a comparison sequence. , recorded as ,in, Represents the reference sequence The Middle M The ordinate value of each, Represents the reference sequence The Middle M Each ordinate value;

[0032] S305. Apply mean normalization to the reference sequence. and comparison sequences The standardization process is performed to obtain the modified reference sequence. and comparison sequences ;

[0033] S306, For each point t Calculate the reference sequence and comparison sequences absolute difference ;

[0034] S307. Determine the minimum difference and maximum value ;

[0035] S308, Based on absolute difference , minimum difference and maximum value Calculate the correlation coefficient;

[0036] S309. Calculate the reference sequence based on the correlation coefficient. and comparison sequences The degree of correlation;

[0037] S3010. Change the translation and rotation matrices, repeat S304-S309, obtain the correlation degree corresponding to each translation and rotation, and take the translation and rotation matrix corresponding to the maximum correlation degree as the optimal translation and rotation matrix. and .

[0038] The beneficial effects of the above-mentioned further scheme are: by using the matrix relationship between the sensor coordinate system and the spatial coordinate system, the two side profiles can be initially aligned, and high registration accuracy can be achieved when the initial positions are close. Fewer iterations can improve the calculation speed, and the registration effect can be evaluated by analyzing the correlation of the overlapping range of the two side profiles.

[0039] Furthermore, the expression for the degree of correlation is as follows:

[0040] ;

[0041] ;

[0042] ;

[0043] in, Indicates the degree of relevance. m This indicates the total number of rail profile data. t This indicates the number of data points for the rail profile. Represents the correlation coefficient. Represents the resolution coefficient. This represents the reference sequence after standardization. The vertical axis value, This represents the reference sequence after standardization. The vertical axis value.

[0044] Furthermore, step S4 includes the following steps:

[0045] Using the optimal translation and rotation matrix and Based on the amount of rail profile data within the overlapping range, the rail profile data is spliced ​​together to obtain the spliced ​​rail profile data.

[0046] The beneficial effects of the above-mentioned further scheme are: given the stable and non-obvious jumps in the data of the overlapping range of rail profiles, the above method makes the calculation simpler and faster, and the spliced ​​rail profile data is more reliable, with high data validity and closer to the true value.

[0047] Furthermore, the splicing process is as follows:

[0048] The changed reference sequence and comparison sequences Recombining the data in ascending order of their horizontal coordinates yields the overlapping range recombination sequence. Initialize the concatenation value , Indicates the first i One rail profile data point;

[0049] Based on splicing value , construct a recursive formula;

[0050] Using the recursive formula, the splicing value of each rail profile data point in the overlapping range is calculated recursively;

[0051] The original rail profile data in the overlapping range is replaced by the calculated splicing value, and the rail profile data outside the original overlapping range is spliced to obtain the measured rail profile data.

[0052] Further, the expression of the recursive formula is as follows:

[0053] ;

[0054] ;

[0055] wherein, represents the first recursive function, represents a smoothing coefficient, represents the first rail profile data point, represents the first recursive function, i , , and , and represent the first, second and third rail profile data points in the recombination sequence , respectively. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 is a schematic diagram of the three-dimensional structure of the device of the present application.

[0057] Figure 2 is a schematic diagram of the three-dimensional structure of the internal data transmission of the device of the present application.

[0058] Figure 3 is a flowchart of the present application.

[0059] wherein, 1-left profile measurement unit, 2-right profile measurement unit, 3-main body walking unit, 4-connection cross beam, 5-assistant walking unit, 6-manual push rod unit, 7-left side limiting unit, 8-right side limiting unit, 9-PC assembly. DETAILED DESCRIPTION

[0060] The specific embodiments of the present application are described below to facilitate understanding of the present application by those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, it is obvious that various changes are within the spirit and scope of the present application as defined and determined by the appended claims, and all inventions utilizing the concept of the present application are within the scope of protection.

[0061] Embodiment 1

[0062] As Figure 2 and Figure 3 The present application provides a steel rail profile dynamic detection device, which comprises a left profile measurement unit 1, a right profile measurement unit 2, a main body walking unit 3, a connecting cross beam 4, an auxiliary walking unit 5, a manual push rod unit 6, a left side limiting unit 7, a right side limiting unit 8 and a PC assembly 9. The left profile measurement unit 1 and the right profile measurement unit 2 are both installed on the main body walking unit 3. The left side limiting unit 7, the right side limiting unit 8 and the PC assembly 9 are all installed on the connecting cross beam 4. The manual push rod unit 6 is installed on the top of the connecting cross beam 4. One end of the connecting cross beam 4 is connected with the auxiliary walking unit 5, and the other end of the connecting cross beam 4 is connected with the main body walking unit 3. Among them, Figure 2 The inside of the trolley is shown, including an encoder assembly and a data acquisition and transmission mainboard.

[0063] In this embodiment, the left profile measurement unit 1 and the right profile measurement unit 2 are fixedly installed on the main body walking unit 3. The left side limiting unit 7 and the right side limiting unit 8 are installed on the connecting cross beam 4. The manual push rod unit 6 is fixedly installed on the top of the connecting cross beam 4. The left end of the connecting cross beam 4 is fixedly connected with the auxiliary walking unit 5, and the right end of the connecting cross beam 4 is fixedly connected with the main body walking unit 3. The PC assembly 9 is fixedly installed on the connecting cross beam 4. The data interaction unit transmits the test data to the PC assembly 9 for processing through the wireless communication module.

[0064] In this embodiment, the data interaction unit can be a data acquisition and transmission circuit board with a network port (receiving cross section data scanned by a line laser sensor), a wireless communication module (wirelessly transmitting data with the PC assembly 9) and a data storage and forwarding function.

[0065] In this embodiment, the working principle of the device is as follows:

[0066] The dynamic detection device is erected on the rail section, the left and right limiting units limit the wheels to tightly adhere to the inner sides of the two rails to prevent the device from swinging left and right during the travel, the detection device is started, the linear laser sensor scans the rail profile data, the cross-section data received is transmitted to the PC component 9 through the wireless communication module in the data interaction unit, the profile data registration splicing method based on the moving average value and the gray correlation degree analysis is used to complete the splicing of the cross-section profile, and thus the rail profile data at the position is obtained. Then, the operator pushes the artificial push rod unit 6, the artificial push rod unit 6 realizes force transmission to the device through the connection with the device structure, drives the dynamic detection device to travel along the rail in the forward direction, and the encoder wheel assembly is triggered to sample the pulse at an interval of 0.5 mm during the travel. Every time the trigger is triggered once, the dynamic detection device collects the rail profile data at the position once. The above steps are repeated until the rail section is collected. The profile data registration splicing method based on the moving average value and the gray correlation degree analysis.

[0067] Embodiment 2

[0068] As Figure 1 shown, the application provides a profile data splicing method, and the implementation method is as follows:

[0069] S1, the rail profile dynamic detection device is erected on the rail section, the limiting wheels of the left limiting unit 7 and the right limiting unit 8 are tightly adhered to the inner sides of the two rails, and the rail profile dynamic detection device is started;

[0070] S2, the artificial push rod is pushed, the dynamic detection device is driven to travel along the rail in the forward direction, the rail profile dynamic detection device is triggered to perform sampling work and scan the rail profile data, and the rail profile data is processed. If the current rail section has not been collected, the artificial push rod is continuously pushed, the dynamic detection device is driven to travel along the rail in the forward direction, and the rail profile dynamic detection device is triggered to continue the sampling work. The implementation method is as follows:

[0071] S201, the artificial push rod is pushed, the dynamic detection device is driven to travel along the rail in the forward direction, and the linear laser sensor is used to scan the rail profile data;

[0072] S202, the moving average method is used to remove the abnormal points of the scanned rail profile data, and the processing of the rail profile data is completed. Specifically,

[0073] Based on the scanned rail profile data, the average value of the previous k-1 points and the current point is calculated for each rail profile data point in the window, the moving average value is obtained, and the previous k-1 points of the rail profile data in the window are set to be null.

[0074] Based on the moving average value, the standard deviation is calculated;

[0075] Calculate the mean of all normalized deviations , standard deviation and threshold value , if the normalized deviation exceeds , it is determined as a jump point, and the jump point is removed, and the data of the rail profile data is completed;

[0076] S3, based on the processed rail profile data, profile registration processing is carried out, and the optimal translation rotation matrix is obtained, and the implementation method is as follows:

[0077] S301, based on the position of the line laser sensor, the matrix relationship between the line laser sensor coordinate system and the space coordinate system is calculated, and the relationship between the line laser sensor measurement image data and the space coordinate system is deduced. Preliminary translation and rotation of the line laser sensor are carried out;

[0078] S302, initialize the translation matrix and the rotation matrix;

[0079] S303, based on the processed rail profile data, the overlapping range of the rail profile data on both sides is retained in the rail profile data obtained by each translation and rotation, wherein the horizontal coordinates of the two rail profile data are the same at this time.

[0080] S304, the longitudinal coordinate value of the non-working edge side coincident rail profile data is taken as the reference sequence , denoted as , and the longitudinal coordinate of the working edge side coincident rail profile data is taken as the comparison sequence , denoted as , wherein represents the longitudinal coordinate value of the i-th in the reference sequence , and M represents the i-th longitudinal coordinate value in the reference sequence . M S305, the reference sequence and the comparison sequence

[0081] are standardized by using mean value, and the changed reference sequence and the comparison sequence are obtained.

[0082] S306, for each point t , the absolute difference value of the reference sequence and the comparison sequence is calculated.

[0083] S307, the minimum difference and the maximum value are determined. ​​

[0084] S308, calculating the correlation coefficient based on the absolute difference value , the minimum difference and the maximum value ;

[0085] S309, based on the correlation coefficient, calculating the correlation degree of the reference sequence and the comparison sequence ;

[0086] S3010, changing the translation matrix and the rotation matrix, repeating S304-S309 to obtain the correlation degree corresponding to each translation and rotation, and taking the translation and rotation matrix corresponding to the maximum correlation degree as the optimal translation and rotation matrix and ;

[0087] S4, splicing the rail profile data through the optimal translation and rotation matrix to obtain the spliced rail profile data, wherein the comparison result of the spliced rail profile data and the standard rail profile data in the database is used to judge the service state of the measured rail, which is specifically:

[0088] According to the amount of rail profile data in the overlapping range and , the rail profile data is spliced to obtain the spliced rail profile data, wherein the splicing process is as follows:

[0089] The changed reference sequence and the comparison sequence are reorganized in the order of small to large horizontal coordinates to obtain the reorganized sequence of the overlapping range , the splicing value is initialized , represents the i th rail profile data point;

[0090] Based on the splicing value , a recursive formula is constructed;

[0091] Using the recursive formula, the splicing value of each rail profile data point in the overlapping range is recursively calculated;

[0092] Using the calculated splicing value to replace the original rail profile data in the overlapping range, and splicing with the rail profile data outside the original overlapping range to obtain the measured rail profile data.

[0093] In this embodiment, since there may be large jump points at the overlapping positions of the detection areas of the line laser sensor, the moving average method is first used to remove outliers from the rail profile data. The window is set to 5 and the threshold is 0.5. The statistical threshold method is used to calculate all deviations within each window and to calculate the standardized deviation within each window. Jump points are determined by the threshold and then the data is cleaned to avoid the subsequent registration algorithm from getting stuck in local optima.

[0094] The specific method for detecting jump points in rail profile data measured by sensors using the moving average method is as follows:

[0095] 1) Calculate the moving average within the window: For each window's rail profile data point Calculate the mean of the previous k-1 points and the current point to obtain the moving average, i.e.:

[0096] ;

[0097] in, Represents the moving average. t This indicates the number of data points for the rail profile. Indicates the first i There are 10 rail profile data points, where k represents a data point.

[0098] Set the first k-1 points of data in the window to NaN to avoid the moving average becoming ineffective.

[0099] 2) Calculate the standardized deviation: ,in, This represents the standard deviation of the data within the window. Indicates the first t One rail profile data point.

[0100] 3) Set thresholds and identify breakpoints: Calculate the mean of all standardized deviations using the statistical thresholding method. and standard deviation and threshold If the deviation exceeds If it is, then it is determined to be a jump point.

[0101] 4) Remove the identified jump points.

[0102] In this embodiment, the profile registration is as follows:

[0103] 1) Calculate the matrix relationship between the line laser sensor coordinate system and the spatial coordinate system based on the position of the line laser sensor, and then deduce the relationship between the line laser sensor measurement image data and the spatial coordinate system, and perform preliminary rotation and translation on the line laser sensor data.

[0104] 2) Initialize the translation matrix and rotation matrix .

[0105] 3) Keep the range of the two rail profile data, which is the same in the horizontal coordinate, and only consider the feature in the vertical coordinate.

[0106] 4) Take the vertical coordinate of the non-working side rail profile data as the reference sequence, denoted as:

[0107] ;

[0108] Take the vertical coordinate of the working side rail profile data as the comparison sequence, denoted as:

[0109] ;

[0110] wherein, is the vertical coordinate value of the i-th point in the reference sequence, is the vertical coordinate value of the i-th point in the comparison sequence. M 5) Standardize the reference sequence and the comparison sequence M using the mean value method, i.e. for each point in the sequence, the transformed

[0111] is: ; wherein,

[0112] is the mean value of the sequence in which the point is located, is the point obtained after standardization of the point

[0113] in the sequence. Further, the changed reference sequence and the comparison sequence are obtained.

[0114] 6) Calculate the absolute difference sequence: for each point , calculate the absolute difference value, i.e. the distance, between the reference sequence and the comparison sequence :

[0115] . 7) Determine the minimum difference

[0116] and the maximum difference .

[0117]

[0118] ​​​​8) Calculate the correlation coefficient:

[0119] ;

[0120] in, This represents the resolution coefficient, which is taken here. .

[0121] 9) Calculate the correlation between the reference sequence and the comparison sequence:

[0122] ;

[0123] in, Indicates the degree of relevance. m This indicates the total number of rail profile data. t This indicates the number of data points for the rail profile. Represents the correlation coefficient. Represents the resolution coefficient. This represents the reference sequence after standardization. The vertical axis value, This represents the reference sequence after standardization. The vertical axis value.

[0124] 10) Change the translation and rotation matrix, repeat steps 4 to 9, and obtain the correlation degree corresponding to each translation and rotation. The translation and rotation matrix corresponding to the maximum correlation degree is the optimal translation and rotation matrix. and .

[0125] In this embodiment, the outline splicing is achieved through the optimal translation and rotation matrix. and The two rail profile data are spliced ​​together. Based on the amount of overlapping data, the following method is used.

[0126] 1) For overlapping recombination sequences Initialize the concatenation value :

[0127] ;

[0128] 2) Establish the recursive formula:

[0129] ;

[0130] in, Indicates the first A recursive function, Represents the smoothing coefficient. Indicates the first i One rail profile data point, Indicates the first A recursive function, , and represent the first, second and third rail profile data points in the recombination sequence .

[0131] 3) Recursively calculate the splicing value of each point;

[0132] 4) Replace the original overlapping range rail profile data with the calculated splicing value, splice with the rail profile data outside the original overlapping range, obtain the measured rail profile data, then compare the rail profile data generated through algorithm processing with the standard profile data in the PC database, record the data and judge the service state of the measured rail according to the data difference.

[0133] In summary, the present application completes rail profile data splicing based on the data processing method of moving average and grey correlation degree analysis, realizes high registration accuracy in the case of close initial position, and judges the registration effect by analyzing the correlation degree of the profile overlapping range on both sides. The algorithm has fast calculation time, good profile data registration effect and high data reliability.

Claims

1. A profile data splicing method applied to a steel rail profile dynamic detection device, the steel rail profile dynamic detection device comprising a left profile measurement unit (1), a right profile measurement unit (2), a main body walking unit (3), a connecting cross beam (4), an auxiliary walking unit (5), a manual push rod unit (6), a left side limiting unit (7), a right side limiting unit (8) and a PC assembly (9); the left profile measurement unit (1) and the right profile measurement unit (2) are both installed on the main body walking unit (3), the left side limiting unit (7), the right side limiting unit (8) and the PC assembly (9) are all installed on the connecting cross beam (4), the manual push rod unit (6) is installed on the connecting cross beam (4), one end of the connecting cross beam (4) is connected with the auxiliary walking unit (5), and the other end of the connecting cross beam (4) is connected with the main body walking unit (3), characterized in that, The method comprises the following steps: S1, erecting the rail profile dynamic detection device on the rail section, and tightly attaching the limiting wheels of the left limiting unit (7) and the right limiting unit (8) to the inner sides of the two rails, and starting the rail profile dynamic detection device; S2, pushing the artificial push rod to drive the dynamic detection device to travel along the rails in the forward direction, triggering the rail profile dynamic detection device to perform sampling work and scan the rail profile data, and processing the rail profile data; if the current rail section has not been completely collected, the artificial push rod is continuously pushed to drive the dynamic detection device to travel along the rails in the forward direction, triggering the rail profile dynamic detection device to continue the sampling work; S3, based on the processed rail profile data, performing profile registration processing to obtain an optimal translation rotation matrix; The S3 comprises the following steps: S301, based on the position of the line laser sensor, calculating the matrix relationship between the line laser sensor coordinate system and the space coordinate system, deducing the relationship between the line laser sensor measurement image data and the space coordinate system, and performing preliminary translation and rotation on the line laser sensor; S302, initializing the translation matrix and the rotation matrix; S303, based on the processed rail profile data, retaining the overlapping range of the rail profile data of the two sides in the rail profile data obtained by each translation and rotation, wherein the horizontal coordinates of the two rail profile data are the same at this time; S304. Use the ordinate values ​​of the non-working side coincident rail profile data as a reference sequence. , recorded as And the ordinate of the rail profile data with overlapping working sides is used as a comparison sequence. , recorded as ,in, Represents the reference sequence The Middle M The ordinate value of each, Represents the reference sequence The Middle M Each ordinate value; S305, normalizing the reference sequence using mean value and the comparison sequence to obtain a changed reference sequence and the comparison sequence ; S306, For each point t Calculate the reference sequence and comparison sequences absolute difference ; S307, determining the minimum difference and the maximum value ; S308、calculating the correlation coefficient based on the absolute difference value , the minimum difference , and the maximum value ​ S309、based on the correlation coefficient, calculate the correlation degree of the reference sequence and the comparison sequence ​ S3010, change the translation matrix and the rotation matrix, repeat S304-S309, obtain the correlation degree corresponding to each translation and rotation, and take the translation and rotation matrix corresponding to the maximum correlation degree as the optimal translation and rotation matrix and ; S4, splicing the rail profile data by using the optimal translation rotation matrix to obtain spliced rail profile data, wherein the comparison result of the spliced rail profile data and the standard rail profile data in the database is used to judge the service state of the measured rail.

2. The profile data splicing method according to claim 1, wherein The S2 comprises the following steps: S201, pushing the artificial push rod to drive the dynamic detection device to travel along the rails in the forward direction, and scanning the rail profile data by using the line laser sensor; S202, removing abnormal points from the scanned rail profile data by using the moving average method, and completing the processing of the rail profile data.

3. The profile data splicing method according to claim 2, wherein The S202 comprises the following steps: Based on the scanned rail profile data, calculating the moving average value of the average of the previous k-1 points and the current point of the rail profile data points in each window, and setting the previous k-1 points of the rail profile data in the window to be null; Based on the moving average value, calculating the standard deviation; The mean of all standard deviations is calculated , the standard deviation and the threshold value are calculated, and if the standard deviation exceeds , the jump is determined and removed, and the data of the rail profile is completed.

4. The profile data splicing method according to claim 3, wherein The expression of the moving average value is as follows: ； in, Represents the moving average. t This indicates the number of data points for the rail profile. Indicates the first i There are 10 rail profile data points, where k represents a data point.

5. The profile data splicing method according to claim 4, wherein The expression of the correlation degree is as follows: ； ； ； in, Indicates the degree of relevance. m This indicates the total number of rail profile data. t This indicates the number of data points for the rail profile. Represents the correlation coefficient. Represents the resolution coefficient. This represents the reference sequence after standardization. The vertical axis value, This represents the reference sequence after standardization. The vertical axis value.

6. The profile data splicing method according to claim 5, wherein The S4 comprises the following steps: By optimal translation rotation matrix And According to the amount of rail profile data in the coincidence range, the rail profile data is spliced to obtain spliced rail profile data.

7. The profile data splicing method according to claim 6, wherein The process of splicing is as follows: The changed reference sequence and comparison sequences Recombining the data in ascending order of their horizontal coordinates yields the overlapping range recombination sequence. Initialize the concatenation value , Indicates the first i One rail profile data point; Based on the stitching value , a recursive formula is constructed; Using a recursive formula, recursively calculating the splicing value of each rail profile data point in the overlapping range; Using the calculated splicing value to replace the original rail profile data in the overlapping range, and splicing the rail profile data outside the original overlapping range to obtain the measured rail profile data.

8. The profile data splicing method according to claim 7, wherein The expression of the recursive formula is as follows: ； ； in, Indicates the first A recursive function, Represents the smoothing coefficient. Indicates the first i One rail profile data point, Indicates the first A recursive function, , and Representing recombination sequences The first, second, and third rail profile data points.

Images