A method for identifying principal points of curved track line shape

Abstract

The present invention provides a method for identifying the principal points of the linear shape of a curved track, comprising: selecting geometric dimension parameters containing linear shape information from the relative measurement data of the track as data samples to be detected according to the signal-to-noise ratio; performing change point detection on the data samples to be detected, and segmenting the data samples to be detected according to the identified change points; performing robust estimation of linear parameters on the segmented data samples, and fitting the slope and intercept of the segmented sample line elements, wherein the intersection of each fitted line element of the segmented sample is the principal point of the curve. The present invention utilizes the characteristic that the pattern or trend of the internal geometric parameters of some tracks will change when they are near the principal point of linear shape control, and obtains an accurate estimate of the principal point of linear shape through change point detection and robust estimation, so as to realize efficient and accurate identification of the principal point of linear shape, and can be used to determine the design value of the linear shape parameters of high-speed railway tracks.

Description

Technical Field

[0001] The present invention relates to the technical field of track detection, and in particular to a method for identifying principal points of a curved track line. Background Art

[0002] The running surface of high-speed rail tracks mainly includes horizontal and vertical straight lines and horizontal and vertical curves. The horizontal line shape includes three types: straight line, circular curve and transition curve, and the vertical line shape includes two types: straight line and circular curve (vertical curve). Curves, turnouts and joints are known as the three major track disease-prone locations.

[0003] Railway curves can generally be divided into front and rear transition curves. 1 , l 2 , curve radius R, steering angle α and other parameters and the corresponding linear control point positions of straight-slow (ZH), slow-round (HY), round-slow (YH), ​​slow-straight (HZ), and track geometric state parameters such as track direction, positive vector, height, level (superelevation), track gauge, track change and twist are used to characterize, measure and control. Due to repetitive load effects and uneven changes in road foundation, the above linear parameters will change.

[0004] To obtain an ideal alignment and a smooth running surface, the geometric dimensions of the track need to be corrected. This process is called track tidying or straightening. The premise of track tidying is to determine the deviation between the actual value of the track geometry and the design value. The geometric design value of the plane curve of the track can be determined by the deflection method, tangent offset method, GPS RTK and coordinate method, and a series of linear main point identification methods are formed. However, the above methods often rely on the measurement of external geometric parameters of the North East High-speed Track, which is suitable for the survey of new lines and the comprehensive maintenance of existing lines, but it is difficult to monitor the changes in the alignment in real time. If the alignment changes cannot be grasped in time, it may lead to deviations in train operation parameters and track tidying plans.

[0005] Track inspection vehicles or track inspection instruments that use the inertial reference method or the midpoint chord measurement method continuously collect the internal geometric parameters of the track along the mileage in a vehicle-mounted manner. Since the measurement of external geometric parameters such as the north-east height is avoided, the measurement efficiency is high. Using the internal geometric parameters such as the track direction, positive vector, height, level (superelevation) to identify the main points of the line shape can achieve timely and efficient estimation of the track line shape. Tao Jie et al. disclosed a curve parameter identification algorithm based on the positive vector diagram and the angle diagram (CN201510632161.7). The identification algorithm uses Hough transform, robust estimation and iterative method to process the positive vector diagram and angle diagram of the track inspection instrument, and achieves a curve main point identification accuracy of ≤0.875m. Other researchers used the curvature data of the track inspection vehicle and proposed to obtain a set of corner points based on the rough solution of the curve feature points extracted using the DP (Douglas-Peucker) algorithm, and connect them in sequence to find the extreme values, and then determine the main point of the curve. This method uses the DP algorithm to quickly calculate the rough solution of the principal point of the curve, but a threshold needs to be set. If the outliers are not effectively eliminated, the accuracy of the principal point estimation will be limited. Summary of the invention

[0006] In view of the above problems, the track direction, verse, height, level (superelevation) and other internal geometric parameters of the track are directly related to the linear parameters, and when near the linear control points such as straight-slow, slow-circular, round-slow, slow-straight, etc., the pattern or trend of the above internal geometric parameters of the track will change. The change point detection method uses certain statistical indicators or statistical methods to observe the state of the internal geometric parameter sequence of the track, and can accurately and effectively estimate the position of the main point of the curve. If the outliers are reasonably eliminated, the main point identification with higher accuracy should be achieved.

[0007] In view of the above situation, it is necessary to provide a method for identifying the principal points of the curved track line shape in order to solve the problems of low efficiency and poor accuracy in identifying the principal points of the track line shape in the prior art.

[0008] A method for identifying principal points of a curved track line shape, comprising:

[0009] According to the signal-to-noise ratio, geometric dimension parameters containing linear information are selected from the track relative measurement data as data samples to be detected;

[0010] Perform change point detection on the data sample to be detected, and segment the data sample to be detected according to the identified change points;

[0011] The linear parameters of the segmented data samples are robustly estimated, and the slope and intercept of the segmented sample line elements are fitted. The intersection of each fitted line element of the segmented sample is the principal point of the curve.

[0012] Furthermore, the above-mentioned method for identifying the principal point of the curved track linear shape, wherein the method for identifying the principal point of the curved track linear shape, is characterized in that the step of selecting geometric dimension parameters containing linear information from the track relative measurement data as the data sample to be detected based on the signal-to-noise ratio comprises:

[0013] Compare the measured values ​​of the geometric dimension parameters in the relative measurement data of the track with the design values ​​determined by the corresponding equipment inventory to obtain the deviation value;

[0014] The design value is taken as the signal and the deviation value is taken as the noise, and the signal-to-noise ratio of each geometric dimension parameter containing linear information is calculated;

[0015] According to the signal-to-noise ratio calculation result, several geometric size parameters containing linear information are selected as data samples to be detected.

[0016] Furthermore, the above-mentioned method for identifying the principal point of the curved track linear shape, wherein the method for identifying the principal point of the curved track linear shape, is characterized in that the step of performing change point detection on the data sample to be detected and segmenting the data sample to be detected according to the identified change points comprises:

[0017] Set the search step factor, jump factor and cost function, and roughly segment the data samples to be tested into several sub-intervals and possible change point sets according to the equipment inventory;

[0018] Perform PELT (Pruned Exact Linear Time) search on the roughly segmented data, and form a possible change point set and the optimal cost path of the data sample to be detected based on the cost function;

[0019] The formed possible change point set is used to accurately segment the data samples to be detected.

[0020] Furthermore, in the above-mentioned method for identifying the principal points of the curved track line, the setting of the search step factor, the jump factor and the cost function is characterized in that the cost function is constructed using the 2-norm of the difference between the data sample to be detected and the linear interpolation result of the corresponding sub-interval.

[0021] Furthermore, the above-mentioned method for identifying principal points of the curved track linearity, wherein the segmented data samples are subjected to robust estimation of linear parameters, is characterized in that the robust estimation method of linear parameters adopted is M estimation.

[0022] Furthermore, in the above-mentioned curved track linear principal point identification method, the track relative measurement data is collected by a level 0 track inspection instrument or by a track inspection vehicle.

[0023] The embodiment of the present invention screens geometric dimension parameters through the signal-to-noise ratio and determines the data samples to be detected, and then constructs a cost function with the 2 norm of the sub-interval residual, accurately segments the data samples to be detected through the PELT search algorithm, and finally uses robust estimation to identify the main points of the curve. It makes full use of the characteristics that the patterns or trends of some internal geometric parameters of the track will change near the main points of linear control, and can realize the automatic identification of the main points of the linear shape, and has strong distinguishing ability and high comprehensive efficiency. The method for identifying the main points of the linear shape of the curved track in this embodiment does not require complicated operations and is suitable for determining the design values ​​of the linear parameters of high-speed railway tracks on site. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a flow chart of a method for identifying principal points of a curved track linear shape in a first embodiment of the present invention;

[0025] Figure 2 is the relative measurement raw data in the first embodiment of the present invention;

[0026] Figure 3 This is the accurate segmentation result in the first embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to facilitate understanding of the present invention, the present invention will be described more fully below with reference to various embodiments, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0029] See also Figure 1 , is a method for identifying the principal point of a curved track line shape in a first embodiment of the present invention, and the detection method is applied to the principal point identification of a railway track to determine the track line shape parameters. The method for identifying the principal point of a curved track line shape includes steps S1 to S3.

[0030] Step S1, selecting geometric dimension parameters containing linear information from track relative measurement data as data samples to be detected according to the signal-to-noise ratio.

[0031] In the existing track linear parameter determination method, the external geometric parameters such as the north-east height coordinates of the curved track are often collected by absolute measurement, and then the linear features are extracted and the coordinate data are segmented through complex functional relationships to extract the linear parameters. The collection of external geometric parameters of the track is a long-lasting and tedious task, while the internal geometric parameters such as track direction, positive vector, height and level (superelevation) are closely related to the track linear shape, and can be efficiently collected by a track inspection vehicle or a track inspection instrument in a relative measurement method. Therefore, it should be considered to identify the main point from the relative measurement data, thereby laying the foundation for the linear parameter extraction. However, due to the existence of track unevenness, it is difficult to accurately identify the main point directly through the internal geometric parameters. The internal geometric parameters should be reasonably selected and the samples to be tested should be determined. In this embodiment, the relative measurement data of the track of the first preset length is collected as the original data. Specifically, the first preset length should contain a curve and at least two different linear shapes. Among them, the relative measurement data of the track can be collected by existing detection equipment, which includes but is not limited to a level 0 track inspection instrument and a track inspection vehicle.

[0032] See also Figure 2 , is the relative measurement raw data in the first embodiment of the present invention. Figure 2 It can be seen that the waveforms of geometric parameters such as left (right) track direction, left (right) height, left (right) verse and horizontal (superelevation) include the front transition curve and some straight lines and circular curves. In order to effectively identify the main point features, certain indicators should be used to screen geometric dimension parameters from the relative measurement raw data as data samples to be detected. The indicator used in the present invention is the signal-to-noise ratio (Signal-Noise Ratio, SNR). Specifically, the measured values ​​of the geometric dimension parameters in the relative measurement data of the track are compared with the design values ​​determined by the corresponding equipment ledger to obtain the deviation values ​​of the geometric dimension parameters; then the design value is used as the signal and the deviation value is used as the noise to calculate the signal-to-noise ratio of each geometric dimension parameter containing linear information, and the calculation method is as follows:

[0033] SNR = 101g (P s / P n )

[0034] Among them, SNR is the signal-to-noise ratio of the geometric size parameter / dB, P s is the average power of the design value, P n Represents the average power of the unevenness.

[0035] According to the signal-to-noise ratio calculation results, several geometric size parameters containing linear information are selected as data samples to be detected. Specifically, Figure 2The horizontal (superelevation) SNR of the sample shown is 49.978 / dB, which is better than other geometric parameters. The horizontal (superelevation) can be selected as the data sample to be detected Y={y(t)}, and the mileage t=[0,1,…,T], where T represents the maximum value of the mileage.

[0036] Step S2, performing change point detection on the data sample to be detected, and segmenting the data sample to be detected according to the identified change points.

[0037] Due to the repetitive load action and uneven changes in the road foundation, as well as the mileage benchmark error and mileage measurement error, the main points of the curve need to be identified on site. To accurately identify the main points of the curve from the data sample Y to be tested, the linear shape to which the data belongs can be segmented according to the characteristics of its pattern or trend change. In this embodiment, the statistical characteristics of the data sample Y to be tested will show a sudden change on both sides of the curve control main point. Using change point detection, the data sample Y to be tested can be effectively segmented.

[0038] Specifically, the search step factor s=1, the jump factor j=1, and the steps of constructing the cost function based on the 2-norm are as follows:

[0039] Step 1: Construct the kth sub-signal y(t k :t k+1 )The cost function of the fitting degree c(y(t k :t k+1 )) is as follows

[0040]

[0041] Where, t k ,t k+1 are the kth and k+1th possible change points respectively; y(t) is the kth sub-signal y(t k :t k+1 ) data, t∈[t k ,t k+1 -1]; α k , β k are the slope and intercept of the linear interpolation of the kth sub-signal respectively.

[0042] Step 2: Construct the objective function V(Ω,y) of the data sample Y to be detected so that the model cost and the minimum are as follows:

[0043]

[0044] Where Ω is a possible set of change points, Ω={t 1 ,t 2 ,…,t K}, t 1 ,t 2 ,tK They are the 1st, 2nd, and Kth possible change points respectively.

[0045] In order to improve the efficiency of searching for change points, the data samples to be detected can be roughly segmented into several sub-intervals and possible change point sets according to the equipment inventory, and a penalty term can be added to the objective function.

[0046] On this basis, a PELT (Pruned Exact Linear Time) search is performed on the roughly segmented data, and the model cost and the minimum optimal cost path and possible change point set are obtained through dynamic programming method.

[0047]

[0048] in, They are The first, second, and Kth elements in .

[0049] Using the formed Treat Y as an accurate segment, the expression is as follows:

[0050]

[0051] See also Figure 3 , which is the accurate segmentation result in the first embodiment of the present invention, Y is divided into three segments according to the mileage: [14.67950, 14.9810], [14.981125, 15.25825] and [15.25950, 15.44750].

[0052] Step S3, robustly estimate the linear parameters of the segmented data samples to obtain the main point mileage of the curve. The slope and intercept of the segmented sample line element are obtained by fitting, and the intersection of each fitting line element of the segmented sample is the main point of the curve.

[0053] Even if the segmentation is accurate, each sub-signal There may still be a few outliers in the distribution, which may cause contamination of the least squares estimate. Under the condition that the samples come from a contaminated distribution, robust estimation can obtain the optimal estimate of the sample by iterative weight adjustment. Classical robust estimation methods include M estimation, L estimation, and R estimation. Among them, M estimation is also called generalized maximum likelihood estimation, which can be used to obtain robust estimation parameters when there are outliers.

[0054] In specific implementation, the sub-signal Perform linear M estimation and fit the sub-signal into a straight line. Obviously, the line elements of the circular curve and the horizontal (high and low) fitting of the clip line are horizontal lines with a slope of about 0, while the front and rear ease curves are oblique lines. At this time, the intersection of each line element is the principal point of the curve.

[0055] Referring to TB / T3147-2020 Railway Track Inspection Instrument, five round trips were actually measured on a certain curved line, and the mileage of the main points such as ZH and HY and the length of the preceding easement curve were estimated as shown in Table 1:

[0056] Table 1

[0057]

[0058] This embodiment uses the signal-to-noise ratio screening level (superelevation) as the data sample to be detected, and then constructs the cost function with the 2 norm of the sub-interval residual, accurately segments the data sample to be detected through the PELT search algorithm, and finally uses robust estimation to identify the main points of the curve. It makes full use of the characteristics that the pattern or trend of some internal geometric parameters of the track will change when they are near the main point of linear control, and can realize the automatic identification of the main point of the linear shape, and has strong distinguishing ability and high comprehensive efficiency. The method for identifying the main point of the linear shape of the curved track in this embodiment does not require complicated operations and is suitable for determining the design values ​​of the linear parameters of high-speed railway tracks on site.

[0059] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0060] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for identifying principal points of a curved track line shape, characterized in that: include: According to the signal-to-noise ratio, geometric dimension parameters containing linear information are selected from the track relative measurement data as data samples to be detected; Perform change point detection on the data sample to be detected, and segment the data sample to be detected according to the identified change points; The segmented data samples are subjected to robust estimation of linear parameters, and the slope and intercept of the segmented sample line elements are obtained by fitting, wherein the intersection of each fitted line element of the segmented sample is the principal point of the curve; The step of selecting geometric dimension parameters containing linear information from the track relative measurement data as data samples to be detected based on the signal-to-noise ratio includes: Compare the measured values ​​of the geometric dimension parameters in the relative measurement data of the track with the design values ​​determined by the corresponding equipment inventory to obtain the deviation value; Taking the design value as the signal and the deviation value as the noise, the signal-to-noise ratio of each geometric dimension parameter containing linear information is calculated. The calculation method is as follows: SNR=101g(P s / P n ) Among them, SNR is the signal-to-noise ratio of the geometric size parameter, P s is the average power of the design value, P n represents the average power of unsmoothness; According to the signal-to-noise ratio calculation result, several geometric size parameters containing linear information are selected as data samples to be detected.

2. The method for identifying the principal point of a curved track line shape according to claim 1, characterized in that: The step of performing change point detection on the data sample to be detected and segmenting the data sample to be detected according to the identified change points comprises: Set the search step factor, jump factor and cost function, and roughly segment the data samples to be tested into several sub-intervals and possible change point sets according to the equipment inventory; Perform PELT search on the roughly segmented data, and form a possible set of change points and the optimal cost path of the data sample to be detected based on the cost function; The formed possible change point set is used to accurately segment the data samples to be detected.

3. The method for identifying principal points of curved track line shape according to claim 2, characterized in that: The cost function is constructed using the 2-norm of the difference between the data sample to be detected and the linear interpolation result of the corresponding sub-interval.

4. The method for identifying principal points of curved track line shape according to claim 1, characterized in that: The segmented data samples are subjected to robust estimation of linear parameters, and the robust estimation method of linear parameters adopted is M estimation.

5. The method for identifying principal points of curved track line shape according to claim 1, characterized in that: The track relative measurement data is collected by a level 0 track inspection instrument or a track inspection vehicle.

Images