A high-speed railway rail profile grinding method, system, equipment and medium

Abstract

The present application belongs to the technical field of rail grinding, and discloses a method, system, equipment and medium for grinding the profile of high-speed railway rails, the method comprising: obtaining an initial data set of the target wheel rail; constructing an equivalent taper prediction model, a vibration characteristic parameter prediction model, a profile deviation value and a grinding scheme prediction model respectively according to the initial data set; predicting the vibration characteristic parameters to be measured by the profile deviation value and the grinding scheme prediction model to obtain the predicted profile deviation value and the predicted grinding scheme, and then obtaining the predicted equivalent taper by the equivalent taper prediction model, and then obtaining the predicted vibration characteristic parameters by the vibration characteristic parameter prediction model; looping in sequence until the predicted vibration characteristic parameters meet the standard parameter values; grinding the target wheel rail according to the predicted vibration characteristic parameters that meet the standard parameter values. The method does not require manual profile measurement, and can obtain grinding schemes for different curves of the line.

Description

Technical Field

[0001] The invention belongs to the technical field of rail grinding, and in particular relates to a method, system, equipment and medium for grinding the profile of a high-speed railway rail. Background Art

[0002] As an important part of modern transportation, the safety and comfort of high-speed railways are directly related to the travel experience of passengers and the efficiency of railway operations. As the core component of high-speed railways, the surface state and geometry of rails are crucial to the running stability and safety of trains. In order to maintain the ideal state of rails, regular grinding operations are essential. Grinding can not only eliminate wear and irregular shapes on the surface of rails, but also effectively extend the service life of rails and reduce maintenance costs. Therefore, how to improve the accuracy and efficiency of rail grinding has become an important topic in the current research of railway maintenance technology.

[0003] At present, the grinding of high-speed railway rails mainly relies on manual operation. The specific process is to first measure the rail profiles of different mileages of high-speed railway lines to obtain their geometric parameters. Then, the measurement results are manually compared with the standard profile to calculate the deviation of each angle. Then, the grinding mode of each profile is formulated based on manual experience.

[0004] However, this process is not only time-consuming and labor-intensive, but also requires high professional skills and experience from the operators. Since the measurement and calculation are all done manually, human errors are prone to occur, resulting in inconsistent angle deviations of the rail profiles at different mileages, thus affecting the grinding effect. Therefore, the existing manual grinding technology, due to the possible different deviations at each measuring point, leads to an inconsistent final grinding pattern, which in turn causes the wave grinding phenomenon on the rail surface after grinding. Summary of the invention

[0005] Based on this, the present application proposes a method, system, equipment and medium for grinding the profile of high-speed railway rails, aiming to solve the problem of existing manual grinding technology, in which the deviation of each measuring point may be different, resulting in the final grinding pattern being not uniform, and thus causing the wave grinding phenomenon on the rail surface after grinding.

[0006] A first aspect of the present application provides a method for grinding a high-speed railway rail profile, the method comprising:

[0007] S1: Acquire an initial data set of the target wheel rail, the initial data set includes rail line grade, rail profile grinding parameters, measured equivalent taper and vibration characteristic parameters, the rail profile grinding parameters include the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding plan;

[0008] S2: constructing an equivalent taper prediction model according to the rail line grade and the rail profile grinding parameters;

[0009] S3: constructing a vibration characteristic parameter prediction model according to the rail line grade and the measured equivalent taper;

[0010] S4: constructing a front-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade and the vibration characteristic parameters, and constructing a back-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade, the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point;

[0011] S5: inputting the vibration characteristic parameters to be measured into the profile deviation value and polishing scheme prediction model to obtain a predicted profile deviation value and a predicted polishing scheme;

[0012] S6: inputting the predicted profile deviation value and the predicted grinding scheme into the equivalent taper prediction model to obtain a predicted equivalent taper;

[0013] S7: inputting the predicted equivalent taper into the vibration characteristic parameter prediction model to obtain predicted vibration characteristic parameters;

[0014] S8: cyclically executing S5 to S8 in sequence until the predicted vibration characteristic parameter meets the standard parameter value;

[0015] S9: Grinding the target wheel rail according to the predicted vibration characteristic parameters that meet standard parameter values.

[0016] Compared with the prior art, the present application provides a method for grinding the profile of high-speed railway rails. The method first obtains an initial data set of the target wheel rail, including the rail line grade, rail profile grinding parameters, measured equivalent taper and vibration characteristic parameters, thereby providing an accurate data basis for the subsequent prediction model construction; an equivalent taper prediction model is constructed according to the rail line grade and rail profile grinding parameters, and the equivalent taper of the rail can be predicted. This technical feature enables the taper change of the rail to be predicted in real time during the rail grinding process; a vibration characteristic parameter prediction model is constructed according to the rail line grade and the measured equivalent taper, which helps to predict the vibration of the rail during operation; the profile grinding amount and the profile deviation value are predicted by constructing a front-end network model and a back-end network model, thereby realizing precise control of the rail grinding process; the front-end network model is based on the rail line grade and the vibration characteristic parameters, and the back-end network model is based on the rail line grade and the vibration characteristic parameters, and the back-end network model is based on the rail line grade and the vibration characteristic parameters. The track grade, measured equivalent taper, marker point coordinates and normal vector deviation angle work together to improve the accuracy of the prediction; the vibration characteristic parameters to be measured are input into the prediction model to obtain the deviation value prediction matrix and the grinding amount prediction matrix, and then the predicted rail profile is obtained; the predicted equivalent taper is obtained according to the equivalent taper prediction model and the predicted rail profile; the predicted vibration characteristic parameters are obtained according to the vibration characteristic parameter prediction model and the predicted equivalent taper. This series of prediction processes realizes real-time monitoring and adjustment of the rail grinding process, ensuring the stability and accuracy of the grinding process; the prediction model is executed cyclically until the predicted vibration characteristic parameters meet the standard parameter values, thus realizing continuous optimization of the rail grinding process; finally, the target wheel rail is ground according to the predicted vibration characteristic parameters that meet the standard parameter values, ensuring that the ground rail meets the use requirements and can effectively avoid the problem of wave grinding on the rail surface after grinding.

[0017] As an optional implementation manner of the first aspect, S1 includes: the initial data set also includes a standard rail profile and a simulated rail profile, wherein: according to the standard rail profile and the simulated rail profile, the x-axis coordinate of the marking point and the deviation angle of the normal vector of the marking point are obtained; according to the x-axis coordinate of the marking point and the deviation angle of the normal vector of the marking point, the rail profile deviation value and the grinding scheme are obtained; according to the rail profile deviation value and the grinding scheme, the rail profile grinding amount is obtained; the left wheel rolling circle radius, the right wheel rolling circle radius and the wheelset lateral displacement of the target wheel rail are obtained to obtain the measured equivalent taper; the vibration acceleration signal is obtained, and according to the signal processing algorithm, the vibration acceleration signal is converted into the vibration characteristic parameter.

[0018] As an optional implementation scheme of the first aspect, S2 includes: constructing a first training data set according to the rail line grade, the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding scheme; normalizing the first training data set to obtain a first standardized matrix; constructing a first objective function for the equivalent taper prediction model, inputting the first standardized matrix into the equivalent taper prediction model and iteratively optimizing the first objective function using a particle swarm algorithm, and finally obtaining the optimized equivalent taper prediction model.

[0019] As an optional implementation of the first aspect, S3 includes: constructing a second training data set based on the rail line grade and the measured equivalent taper; normalizing the second training data set to obtain a second standardized matrix; constructing a second objective function for the vibration characteristic parameter prediction model, inputting the second standardized matrix into the vibration characteristic parameter prediction model and iteratively optimizing the second objective function using a particle swarm algorithm, and finally obtaining the optimized characteristic parameter prediction model.

[0020] As an optional implementation of the first aspect, S4 includes: constructing a third training data set according to the rail line grade and the vibration characteristic parameters; normalizing the third training data set to obtain a third standardized matrix; constructing a third objective function for the front-end network model, inputting the third standardized matrix into the front-end network model and iteratively optimizing the third objective function using a particle swarm algorithm, and finally obtaining the optimized front-end network model; constructing a fourth training data set according to the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point; normalizing the fourth training data set to obtain a fourth standardized matrix; constructing a fourth objective function for the back-end network model, inputting the fourth standardized matrix into the back-end network model and iteratively optimizing the fourth objective function using a particle swarm algorithm, and finally obtaining the optimized back-end network model; combining the optimized front-end network model and the optimized back-end network model into the profile deviation value and grinding scheme prediction model.

[0021] As an optional implementation of the first aspect, the S2 further includes: according to the wheel-rail contact geometry theory, calculating the equivalent taper of the rail profile at different lateral displacements and recording the critical value of the equivalent taper, the critical value of the equivalent taper is used to judge whether the grinding is qualified according to the predicted equivalent taper of the wheel-rail. Specifically, i normal changes are set at each marking point to obtain the normal deviation value of the simulated rail profile and the standard rail profile, and the normal deviation value is equal to the size of the simulated rail profile grinding amount; the simulated rail profile grinding amount is arranged and combined to obtain The simulated rail profiles of various combinations are calculated; the equivalent tapers λ1, λ2, ..., λ of the simulated rail profiles under different lateral displacements are calculated for each combination. j And record the critical value λ of the equivalent taper min and λ max .

[0022] As an optional implementation of the first aspect, the S3 also includes: calculating the vibration acceleration signal of the rail profile under different lateral displacements through wheel dynamics simulation software and recording the maximum lateral vibration acceleration value and the maximum vertical vibration acceleration value as critical values ​​for judging the vibration characteristic parameters, wherein the critical value is used to judge whether the grinding is qualified according to the predicted vibration characteristic parameters of the wheel and rail. Specifically, the measured rail profile and the corresponding measured equivalent taper under different lateral displacements are introduced into the dynamics simulation software to calculate the corresponding vibration acceleration signal; according to the vibration acceleration signal, the corresponding maximum lateral vibration acceleration value a is recorded. xmax and the maximum vertical vibration acceleration a ymax .

[0023] A second aspect of the present application provides a high-speed railway rail profile grinding system, the system comprising:

[0024] The data acquisition module includes S1: acquiring an initial data set of a target wheel rail, wherein the initial data set includes rail line grade, rail profile grinding parameters, measured equivalent taper, and vibration characteristic parameters, wherein the rail profile grinding parameters include x-axis coordinates of a marking point, a normal vector deviation angle of a marking point, a rail profile deviation value, a rail profile grinding amount, and a grinding plan;

[0025] The model building module includes S2: building an equivalent taper prediction model according to the rail line grade and the rail profile grinding parameters; S3: building a vibration characteristic parameter prediction model according to the rail line grade and the measured equivalent taper; S4: building a front-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade and the vibration characteristic parameters, and building a back-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade, the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point;

[0026] The prediction module includes S5: inputting the vibration characteristic parameter to be measured into the profile deviation value and the grinding scheme prediction model to obtain the predicted profile deviation value and the predicted grinding scheme; S6: inputting the predicted profile deviation value and the predicted grinding scheme into the equivalent taper prediction model to obtain the predicted equivalent taper; S7: inputting the predicted equivalent taper into the vibration characteristic parameter prediction model to obtain the predicted vibration characteristic parameter;

[0027] The judgment and grinding module includes S8: executing S5 to S8 in a loop in sequence until the predicted vibration characteristic parameters meet the standard parameter values; S9: grinding the target wheel rail according to the predicted vibration characteristic parameters that meet the standard parameter values.

[0028] The third aspect of the present application provides an electronic device, comprising: a processor; a memory for storing executable instructions of the processor; wherein the processor is configured to execute the executable instructions to implement the above-mentioned high-speed railway rail profile grinding method.

[0029] A fourth aspect of the present application provides a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the above-mentioned high-speed railway rail profile grinding method.

[0030] Additional aspects and advantages of the present application will be given in part in the following description, and in part will become apparent from the following description, or will be understood through the embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flow chart of a high-speed railway rail profile grinding method proposed in the first embodiment of the present application;

[0032] Figure 2 This is a design outline diagram of the rail grinding in the first embodiment of the present application;

[0033] Figure 3 This is a structural diagram of the first equivalent taper prediction model in the first embodiment of the present application;

[0034] Figure 4 A structural diagram of a vibration characteristic parameter prediction model in the first embodiment of the present application;

[0035] Figure 5 It is a structural diagram of the profile deviation value and polishing scheme prediction model in the first embodiment of the present application;

[0036] Figure 6 This is a schematic structural diagram of a high-speed railway rail profile grinding system proposed in the second embodiment of the present application.

[0037] The following specific implementation methods will further illustrate the present application in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here. In addition, the "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.

[0040] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.

[0041] Example 1

[0042] See also Figure 1 , is a flow chart of a high-speed railway rail profile grinding method proposed in the first embodiment of the present application, wherein the proposed method includes the following S1 to S9:

[0043] S1: Obtain an initial data set of the target wheel rail, wherein the initial data set includes rail line grade, rail profile grinding parameters, measured equivalent taper and vibration characteristic parameters, wherein the rail profile grinding parameters include the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding plan.

[0044] It should be noted that the initial data set also includes the standard rail profile and the simulated rail profile, where:

[0045] According to the standard rail profile and the simulated rail profile, the x-axis coordinate of the marking point and the deviation angle of the normal vector of the marking point are obtained; according to the x-axis coordinate of the marking point and the deviation angle of the normal vector of the marking point, the rail profile deviation value and the grinding scheme are obtained; according to the rail profile deviation value and the grinding scheme, the rail profile grinding amount is obtained;

[0046] Obtaining the left wheel rolling circle radius, the right wheel rolling circle radius and the wheel pair lateral displacement of the target wheel rail to obtain the measured equivalent taper;

[0047] A vibration acceleration signal is acquired, and the vibration acceleration signal is converted into the vibration characteristic parameter according to a signal processing algorithm.

[0048] Exemplarily, first, the initial data set of the target wheel rail can be obtained through field measurement or equipment collection. This data set includes the rail line grade, which is usually used to describe the technical standards and transportation capacity of the railway line, including the division according to the designed driving speed of the line; rail profile grinding parameters, such as the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding scheme number, which reflect the geometry and maintenance status of the rail; the standard profile and simulated profile of the rail, which are important references for evaluating the actual rail status. Specifically, a large number of initial data sets of on-site target wheel rails are obtained, and according to the "Management Measures for High-speed Railway Rail Grinding", n marking points are selected on the inside and outside of the profile, and the corresponding coordinates are (x1, y1), (x2, y2)...(x n ,y n ), the normal vector angle corresponding to each marked point is recorded as ɑ1, ɑ2, ...., ɑ n ,like Figure 2 The figure shows the design profile of rail grinding. Rail grinding needs to be performed along the normal vector at n marking points corresponding to the rail profile and the standard profile. The grinding amount at each marking point is recorded as d1, d2, ..., d n The grinding amount determines where and how much grinding is performed. The matching grinding amount calculated according to the corresponding grinding plan can recommend the optimal grinding strategy for the current rail state.

[0049] Secondly, collect the rolling circle radius of the left and right wheels of the target wheel rail, as well as the lateral displacement of the wheelset. Using these data, combined with the vehicle dynamics model, the measured equivalent conicity can be calculated, which is an important indicator for evaluating track smoothness. Specifically, the calculation formula for the measured equivalent conicity is: ,in, represents the measured equivalent taper, represents the radius of the left wheel rolling circle, Indicates the radius of the right wheel rolling circle, Indicates the lateral displacement of the wheelset.

[0050] Finally, during the operation, the vibration acceleration signal is collected by the sensor installed on the wheel. Then, the signal processing algorithm (such as Fourier transform, wavelet analysis, etc.) is applied to convert the original vibration acceleration signal into a series of vibration characteristic parameters, such as frequency spectrum, root mean square value, peak value, etc. These parameters can reflect the vehicle vibration caused by track unevenness.

[0051] S2: constructing an equivalent taper prediction model according to the rail line grade and the rail profile grinding parameters.

[0052] In some embodiments, this step includes: constructing a first training data set based on the rail line grade, the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding scheme; normalizing the first training data set to obtain a first standardized matrix; constructing a first objective function for the equivalent taper prediction model, inputting the first standardized matrix into the equivalent taper prediction model and using a particle swarm algorithm to iteratively optimize the first objective function, and finally obtaining the optimized equivalent taper prediction model.

[0053] First, the six basic data of rail line grade, x-axis coordinate of marking point, normal deviation angle of marking point, rail profile deviation value, rail profile grinding amount and grinding scheme are constructed into a first training data set. Exemplarily, the basic data of the first training data set is represented in the following form:

[0054] XLDJ=[DJ1,DJ2,DJ3,DJ4,DJ5,DJ6,DJ7,DJ8,DJ9,......,DJ n ]

[0055] DMFN=[fn1,fn2,fn3,fn4,fn5,fn6,fn7,fn8,fn9,....,fn n ]

[0056] PC=[kx pc1 ,kx pc2 ,kx pc3 ,kx pc4 ,kx pc5 ,kx pc6 ,kx pc7 ,kx pc8 ,kx pc9 ,....,kx pcn ]

[0057] DM=[kx dm1 ,kxdm2 ,kx dm3 ,kx dm4 ,kx dm5 ,kx dm6 ,kx dm7 ,kx dm8 ,kx dm9 ,....,kx dmn ]

[0058] FXPCJD=[α1,α2,α3,α4,α5,α6,α7,α8,α9,α 10 ,α 11 ,α 12 ,α 13 ,α 14 ,α 15 ,α 16 ,α 17 ,α 18 ,α 19 ,α 20 ]

[0059] BJD=[x1,x2,x3,x4,x5,x6,x7,x8,x9,x 10 ,x 11 ,x 12 ,x 13 ,x 14 ,x 15 ,x 16 ,x 17 ,x 18 ,x 19 ,x 20 ]

[0060] Where XLDJ represents the rail line grade, DMFN represents the grinding scheme, PC represents the rail profile deviation value vector, DM represents the rail profile grinding amount vector, FXPCJD represents the deviation angle of the normal vector of the marking point, BJD represents the x-axis coordinate of the marking point, and kx pc Indicates the deviation value of each marking point, kx dm Indicates the grinding amount for each marked point.

[0061] Secondly, the above database basic matrix (i.e., the first training data set) is normalized to obtain a first standardized matrix. According to the first standardized matrix, the input matrix H1 of the neural network model is constructed.

[0062] H1=[XLDJ,DMFA,PC,DM,FZPCJD,BJD]

[0063] Where H1 represents the input matrix of the neural network, XLDJ represents the rail line grade, DMFN represents the grinding scheme, PC represents the rail profile deviation value vector, DM represents the rail profile grinding amount vector, FZPCJD represents the normal vector deviation angle of the marking point, and BJD represents the x-axis coordinate of the marking point.

[0064] Subsequently, the input matrix H1 is processed by a neural network algorithm to establish an equivalent taper prediction model, which is convenient for distinguishing from the front-end network model in S4 below. The equivalent taper prediction model here is called the first equivalent taper prediction model. The structure of the first equivalent taper prediction model is as follows: Figure 3 As shown. The output matrix is:

[0065] DXZD1=W Z1 *tansig(W f1 *H1+b f1 )+b Z1

[0066] Where DXZD1 represents the predicted first equivalent conicity result matrix of the wheel, W Z1 Represents the total weight coefficient matrix, W fl represents the component weight coefficient matrix, H1 represents the input matrix of the neural network, b fl represents the component bias coefficient, b Z1 Represents the total bias coefficient.

[0067] Finally, the first objective function of the first equivalent taper prediction model is constructed:

[0068] ,

[0069] Where, DXZD1 JD represents the error norm of the wheel first equivalent taper prediction result, i represents the i-th row of the wheel first equivalent taper prediction result matrix, j represents the j-th column of the wheel first equivalent taper prediction result matrix, m represents the number of rows of the wheel first equivalent taper prediction result matrix, and n represents the number of columns of the wheel first equivalent taper prediction result matrix.

[0070] The first standardized matrix is ​​input into the first equivalent taper prediction model and the first objective function is iteratively optimized using a particle swarm algorithm, and the first equivalent taper prediction model is tested using a test data set to finally obtain an optimized first equivalent taper prediction model.

[0071] In some embodiments, step S2 further includes: according to the wheel-rail contact geometry theory, calculating the equivalent taper of the rail profile at different lateral displacements and recording the critical value of the equivalent taper, wherein the critical value of the equivalent taper is used as one of the criteria for judging whether the grinding is qualified according to the predicted equivalent taper of the wheel and rail (preparing for the execution of the loop judgment in S6 below), specifically: according to the designed driving speed of the rails of different line grades, setting i normal changes at each marking point according to Table 1 (for example: Figure 2 As shown in the figure, the x-axis coordinate of the mark point 1 is within the range of -25mm to +25mm in the horizontal direction of the rail head. If the design speed of the rail is less than 200km / h, the deviation value range is +0.3 / -0.3. The range is divided into i-1 equal parts to obtain i different normal changes from -0.3 to +0.3), thereby obtaining i deviation values ​​of the 20 mark points. Assume that the i deviation values ​​of the 20 mark points are arranged and combined to obtain The simulated rail profiles of the combinations and the normal deviations between the simulated rail profiles and the standard rail profiles are calculated; the equivalent tapers λ1, λ2, ..., λ of the simulated rail profiles under different lateral displacements are calculated for each combination. j And record the critical value λ of the equivalent taper min and λ max .

[0072] Table 1 Different rail head lateral range deviation values ​​corresponding to different design driving speeds

[0073]

[0074] S3: Constructing a vibration characteristic parameter prediction model according to the rail line grade and the measured equivalent taper.

[0075] In some embodiments, this step includes: constructing a second training data set based on the rail line grade and the measured equivalent taper; normalizing the second training data set to obtain a second standardized matrix; constructing a second objective function for the vibration characteristic parameter prediction model, inputting the second standardized matrix into the vibration characteristic parameter prediction model and using a particle swarm algorithm to iteratively optimize the second objective function, and finally obtaining the optimized characteristic parameter prediction model.

[0076] First, the two types of basic data, namely, the obtained rail line grade and the measured equivalent taper, are constructed into a second training data set. Exemplarily, the basic data of the second training data set is represented in the following form:

[0077] XLDJ=[DJ1,DJ2,DJ3,DJ4,DJ5,DJ6,DJ7,DJ8,DJ9,......,DJ n ]

[0078] DXZD=[zd1,zd2,zd3,zd4,zd5,zd6,zd7,zd8,zd9,....,zd n ]

[0079] Where XLDJ represents the rail line grade, DXZD represents the measured equivalent taper vector, and zd represents the measured equivalent taper at each lateral displacement.

[0080] Secondly, the second training data set is normalized to obtain a second standardized matrix, and the input matrix H2 of the neural network model is constructed based on the second standardized matrix.

[0081] H2=[XLDJ,DXZD]

[0082] Where XLDJ represents the rail line grade, and DXZD represents the measured equivalent taper vector.

[0083] Subsequently, the input matrix H2 is processed by a neural network algorithm to construct a vibration characteristic parameter prediction model. The structure of the vibration characteristic parameter prediction model is as follows: Figure 4 As shown. The output matrix is:

[0084] ZDTZ=W Z2 *tansig(W f2 *H2+b f2 )+b Z2

[0085] Where ZDTZ represents the vibration characteristic parameter vector, W Z2 Represents the total weight coefficient matrix, W f2 represents the component weight coefficient matrix, H2 represents the input matrix of the neural network, b f2 represents the component bias coefficient, b Z2 Represents the total bias coefficient.

[0086] Finally, the first objective function of the above vibration characteristic parameter prediction model is constructed:

[0087] ,

[0088] Where ZDTZ JD represents the error norm of the vibration characteristic parameter prediction result, i represents the i-th row of the wheel equivalent taper prediction result matrix, j represents the j-th column of the wheel equivalent taper prediction result matrix, m represents the number of rows of the wheel equivalent taper prediction result matrix, and n represents the number of columns of the wheel equivalent taper prediction result matrix.

[0089] The second standardized matrix is ​​input into the vibration characteristic parameter prediction model and the second objective function is iteratively optimized using the particle swarm algorithm, and the vibration characteristic parameter prediction model is tested using a test data set to finally obtain an optimized vibration characteristic parameter prediction model.

[0090] In some embodiments, step S3 further includes: calculating the vibration acceleration signal of the rail profile under different lateral displacements through wheel dynamics simulation software and recording the critical value of the vibration characteristic parameter, wherein the critical value of the vibration characteristic parameter is used to determine whether the grinding is qualified according to the predicted vibration characteristic parameter of the wheel rail (preparing for the execution of the loop judgment in S7 below), specifically: importing the measured rail profile and the corresponding measured equivalent taper under different lateral displacements into the dynamics simulation software, calculating the corresponding vibration acceleration signal; recording the corresponding maximum lateral vibration acceleration value a according to the vibration acceleration signal; xmax and the maximum vertical vibration acceleration a ymax It should be noted that a xmax Corresponding to the lateral characteristic parameter amplitude HXFZ in the vibration characteristic parameter, a ymax Corresponding to the vertical characteristic parameter amplitude CXFZ in the vibration characteristic parameters.

[0091] S4: A front-end network model of a profile deviation value and a grinding scheme prediction model is constructed according to the rail line grade and the vibration characteristic parameters, and a back-end network model of a profile deviation value and a grinding scheme prediction model is constructed according to the rail line grade, the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point.

[0092] In some embodiments, step S4 includes:

[0093] A third training data set is constructed according to the rail line grade and the vibration characteristic parameters; the third training data set is normalized to obtain a third standardized matrix; a third objective function of the front-end network model is constructed, the third standardized matrix is ​​input into the front-end network model and the third objective function is iteratively optimized using a particle swarm algorithm, and finally the optimized front-end network model is obtained;

[0094] Constructing a fourth training data set according to the measured equivalent taper, the x-axis coordinates of the marking points, and the deviation angle of the normal vector of the marking points; normalizing the fourth training data set to obtain a fourth standardized matrix; constructing a fourth objective function for the back-end network model, inputting the fourth standardized matrix into the back-end network model, and iteratively optimizing the fourth objective function using a particle swarm algorithm, and finally obtaining the optimized back-end network model;

[0095] The optimized front-end network model and the optimized back-end network model are combined into the profile deviation value and polishing scheme prediction model.

[0096] First, the two types of basic data, namely, the rail line grade and the vibration characteristic parameters, are constructed as a third training data set. For example, the basic data in the third training data set are represented in the following form:

[0097] XLDJ=[DJ1,DJ2,DJ3,DJ4,DJ5,DJ6,DJ7,DJ8,DJ9,......,DJ n ]

[0098] ZDTZ=[HXZD,CXZD]

[0099] HXZD=[HXSYTZ,HXPYTZ,HXQTTZ]

[0100] CXZD=[CXSYTZ,CXPYTZ,CXQTTZ]

[0101] Where XLDJ represents the rail line grade, ZDTZ represents the vibration characteristic parameter vector, HXZD represents the lateral vibration characteristic parameter vector, and CXZD represents the vertical vibration characteristic parameter vector.

[0102] Among them, the lateral vibration characteristic parameter vector HXZD includes the lateral vibration time domain characteristic parameter vector HXSYTZ, the lateral vibration frequency domain characteristic parameter vector HXPYTZ and the lateral vibration other characteristic parameter vector HXQTTZ, and the vertical vibration characteristic parameter vector CXZD includes the vertical vibration time domain characteristic parameter vector CXSYTZ, the vertical vibration time domain characteristic parameter vector CXPYTZ and the vertical vibration other characteristic parameter vector CXQTTZ, which are respectively represented by the following matrix forms:

[0103] HXSYTZ=[HXFZ,HXJZ,HXFC]

[0104] HXPYTZ=[HXPSD,HXZP]

[0105] HXQTTZ=[HXYD,HXPD]

[0106] CXSYTZ=[CXFZ,CXJZ,CXFC]

[0107] CXPYTZ=[CXPSD,CXZP]

[0108] CXQTTZ=[CXYD,CXPD]

[0109] Wherein, HXFZ represents the lateral vibration amplitude, HXJZ represents the lateral vibration mean, HXFC represents the lateral vibration variance, HXPSD represents the lateral vibration power spectral density (PSD), HXZP represents the lateral vibration main frequency, HXYD represents the lateral vibration margin, HXPD represents the lateral vibration skewness, CXFZ represents the vertical vibration amplitude, CXJZ represents the vertical vibration mean, CXFC represents the vertical vibration variance, CXPSD represents the vertical vibration power spectral density (PSD), CXZP represents the vertical vibration main frequency, CXYD represents the vertical vibration margin, and CXPD represents the vertical vibration skewness.

[0110] Secondly, the third training data set is normalized to obtain a third standardized matrix. According to the third standardized matrix, an input matrix H3 of the neural network model is constructed.

[0111] H3=[XLDJ,ZDTZ]

[0112] Where XLDJ represents the rail line grade, and ZDTZ represents the vibration characteristic parameter vector.

[0113] Subsequently, the input matrix H3 is processed by the neural network algorithm to establish the front-end network model of the profile deviation value and the polishing scheme prediction model. Since a prediction model for the equivalent taper also needs to be constructed here, it is convenient to distinguish it from the first equivalent taper prediction model in S2 above. The front-end network model here is called the second equivalent taper prediction model, and the output matrix is:

[0114] DXZD2=W Z3 *tansig(W f3 *H3+b f3 )+b Z3

[0115] Where DXZD2 represents the second equivalent cone prediction result matrix, W Z3 Represents the total weight coefficient matrix, W f3 represents the component weight coefficient matrix, H3 represents the input matrix of the neural network, b f3 represents the component bias coefficient, b Z3 Represents the total bias coefficient.

[0116] Finally, the third objective function of the above second equivalent taper prediction model is constructed: ,

[0117] Where, DXZD2 JDrepresents the error norm of the wheel second equivalent taper prediction result, i represents the i-th row of the wheel second equivalent taper prediction result matrix, j represents the j-th column of the wheel second equivalent taper prediction result matrix, m represents the number of rows of the wheel second equivalent taper prediction result matrix, and n represents the number of columns of the wheel second equivalent taper prediction result matrix.

[0118] The third standardized matrix is ​​input into the second equivalent taper prediction model and the third objective function is iteratively optimized using the particle swarm algorithm, and the above-mentioned second equivalent taper prediction model is tested with a test data set, and finally an optimized second equivalent taper prediction model is obtained, that is, an optimized front-end network model.

[0119] It should be noted that, while constructing the second equivalent taper prediction model (the front-end network model of the profile deviation value and polishing scheme prediction model), the back-end network of the profile deviation value and polishing scheme prediction model is also constructed. The front-end network and the back-end network are combined into a complete profile deviation value and polishing scheme prediction model, such as Figure 5 The construction process of the back-end network of the profile deviation value and polishing scheme prediction model is as follows:

[0120] First, the four basic data of the obtained rail line grade, measured equivalent taper, x-axis coordinate of the marking point and normal vector deviation angle of the marking point are constructed into a fourth training data set. Exemplarily, the basic data of the fourth training data set is represented in the following form:

[0121] XLDJ=[DJ1,DJ2,DJ3,DJ4,DJ5,DJ6,DJ7,DJ8,DJ9,......,DJ n ]

[0122] DXZD=[zd1,zd2,zd3,zd4,zd5,zd6,zd7,zd8,zd9,....,zd n ]

[0123] FXPCJD=[α1,α2,α3,α4,α5,α6,α7,α8,α9,α 10 ,α 11 ,α 12 ,α 13 ,α 14 ,α 15 ,α 16 ,α 17 ,α 18 ,α 19 ,α 20 ]

[0124] BJD=[x1,x2,x3,x4,x5, x6 ,x7,x8,x9,x 10,x 11 ,x 12 ,x 13 ,x 14 ,x 15 ,x 16 ,x 17 ,x 18 ,x 19 ,x 20 ]

[0125] Where XLDJ represents the rail line grade, DXZD represents the measured equivalent taper vector, zd represents the equivalent taper at each lateral displacement, FXPCJD represents the normal vector deviation angle of the marking point, and BJD represents the x-axis coordinate of the marking point.

[0126] Secondly, the database basic matrix (ie, the fourth training data set) is normalized to obtain a fourth standardized matrix. Based on the fourth standardized matrix, an input matrix H4 of the neural network model is constructed.

[0127] H4=[XLDJ,DXZD,FXPCJD,BJD]

[0128] Where XLDJ represents the rail line grade, DXZD represents the measured equivalent taper vector, FXPCJD represents the normal vector deviation angle of the marking point, and BJD represents the x-axis coordinate of the marking point.

[0129] Subsequently, the input matrix H4 is processed using a neural network algorithm to establish a back-end network model. The back-end network model has two output terminals, which are used to output the grinding scheme prediction result matrix and the rail deviation value prediction result matrix, respectively, which are expressed in matrices as follows:

[0130] DMFN=W Z4 *tansig(W f4 *H4+b f4 )+b Z4 ,

[0131] PC=W Z5 *tansig(W f5 *H4+b f5 )+b Z5

[0132] Where DMFN represents the grinding scheme prediction result matrix, PC represents the rail deviation value prediction result matrix, and W Z4 and W Z5 Represents the total weight coefficient matrix, W f4 and W f5 represents the component weight coefficient matrix, H4 represents the input matrix of the neural network, b f4 and b f5 represents the component bias coefficient, bZ4 and b Z5 Represents the total bias coefficient.

[0133] Finally, the objective functions of the two output ends (grinding scheme and rail deviation value) of the above back-end network model are constructed respectively:

[0134] ,

[0135] Where, DMFN JD Represents the error norm of the polishing plan prediction result, PC JD It represents the error norm of the rail deviation value prediction result, i represents the i-th row of the back-end network model prediction result matrix, j represents the j-th column of the back-end network model prediction result matrix, m represents the number of rows of the back-end network model prediction result matrix, and n represents the number of columns of the back-end network model prediction result matrix.

[0136] The fourth standardized matrix is ​​input into the backend network model and the particle swarm algorithm is used to iteratively optimize the two output ends in the fourth objective function respectively, and the backend network model is tested with a test data set to finally obtain an optimized backend network model.

[0137] Combine the above optimized front-end network model and the optimized back-end network model to perform model splicing to form the final profile deviation value and polishing plan prediction model.

[0138] S5: inputting the vibration characteristic parameters to be measured into the profile deviation value and polishing scheme prediction model to obtain a predicted profile deviation value and a predicted polishing scheme;

[0139] S6: inputting the predicted profile deviation value and the predicted grinding scheme into the equivalent taper prediction model to obtain a predicted equivalent taper;

[0140] It should be noted that it is necessary to determine whether to execute the loop step according to the set critical value of the equivalent taper (refer to the critical value of the equivalent taper set in the implementation mode proposed in the above S2). If the predicted equivalent taper exceeds the set critical value range of the equivalent taper, the loop step needs to be executed, otherwise, the loop step does not need to be executed.

[0141] S7: inputting the predicted equivalent taper into the vibration characteristic parameter prediction model to obtain predicted vibration characteristic parameters;

[0142] It should be noted that it is necessary to determine whether to execute the loop step according to the critical value of the set vibration characteristic parameter (refer to the critical value of the vibration characteristic parameter set in the embodiment proposed in S3 above). If the predicted vibration characteristic parameter exceeds the critical value range of the set vibration characteristic parameter, the loop step needs to be executed, otherwise, the loop step does not need to be executed.

[0143] S8: cyclically executing S5 to S8 in sequence until the predicted vibration characteristic parameter meets the standard parameter value;

[0144] S9: Grinding the target wheel rail according to the predicted vibration characteristic parameters that meet standard parameter values.

[0145] In summary, this method first obtains the initial data set of the target wheel rail, including rail line grade, rail profile grinding parameters, measured equivalent taper and vibration characteristic parameters, to provide an accurate data basis for the subsequent prediction model construction; an equivalent taper prediction model is constructed according to the rail line grade and rail profile grinding parameters, which can predict the equivalent taper of the rail. This technical feature enables the real-time prediction of the taper change of the rail during the rail grinding process; a vibration characteristic parameter prediction model is constructed according to the rail line grade and the measured equivalent taper, which helps to predict the vibration of the rail during operation; the profile grinding amount and profile deviation value are predicted by constructing the front-end network model and the back-end network model, thereby realizing the precise control of the rail grinding process; the front-end network model is based on the rail line grade and vibration characteristic parameters, and the back-end network model is based on the rail line grade, the measured equivalent taper, The coordinates of the marking points and the normal vector deviation angle work together to improve the accuracy of the prediction; the vibration characteristic parameters to be measured are input into the prediction model to obtain the deviation value prediction matrix and the grinding amount prediction matrix, and then the predicted rail profile is obtained; according to the equivalent taper prediction model and the predicted rail profile, the predicted equivalent taper is obtained; and according to the vibration characteristic parameter prediction model and the predicted equivalent taper, the predicted vibration characteristic parameters are obtained. This series of prediction processes realizes real-time monitoring and adjustment of the rail grinding process, ensuring the stability and accuracy of the grinding process; by cyclically executing the prediction model until the predicted vibration characteristic parameters meet the standard parameter values, the continuous optimization of the rail grinding process is realized; finally, the target wheel rail is ground according to the predicted vibration characteristic parameters that meet the standard parameter values, ensuring that the ground rail meets the use requirements and can effectively avoid the problem of wave grinding on the rail surface after grinding.

[0146] Example 2

[0147] See also Figure 6 , which is a schematic structural diagram of a high-speed railway rail profile grinding system proposed in the second embodiment of the present application, the system comprises:

[0148] The data acquisition module 10 includes S1: acquiring an initial data set of a target wheel rail, wherein the initial data set includes a rail line grade, a rail profile grinding parameter, a measured equivalent taper, and a vibration characteristic parameter, wherein the rail profile grinding parameter includes an x-axis coordinate of a marking point, a normal vector deviation angle of a marking point, a rail profile deviation value, a rail profile grinding amount, and a grinding plan;

[0149] The model building module 20 includes S2: building an equivalent taper prediction model according to the rail line grade and the rail profile grinding parameters; S3: building a vibration characteristic parameter prediction model according to the rail line grade and the measured equivalent taper; S4: building a front-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade and the vibration characteristic parameters, and building a back-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade, the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point;

[0150] The prediction module 30 includes S5: inputting the vibration characteristic parameter to be measured into the profile deviation value and the grinding scheme prediction model to obtain a predicted profile deviation value and a predicted grinding scheme; S6: inputting the predicted profile deviation value and the predicted grinding scheme into the equivalent taper prediction model to obtain a predicted equivalent taper; S7: inputting the predicted equivalent taper into the vibration characteristic parameter prediction model to obtain a predicted vibration characteristic parameter;

[0151] The judgment and grinding module 40 includes S8: executing S5 to S8 in a loop in sequence until the predicted vibration characteristic parameter meets the standard parameter value; S9: grinding the target wheel rail according to the predicted vibration characteristic parameter that meets the standard parameter value.

[0152] On the other hand, the present application also proposes an electronic device, comprising: a processor; a memory for storing executable instructions of the processor; wherein the processor is configured to execute the executable instructions to implement the above-mentioned high-speed railway rail profile grinding method.

[0153] On the other hand, the present application also proposes a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the above-mentioned high-speed railway rail profile grinding method.

[0154] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0155] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0156] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A method for grinding the profile of a high-speed railway rail, characterized in that: The method comprises: S1: Acquire an initial data set of the target wheel rail, the initial data set includes rail line grade, rail profile grinding parameters, measured equivalent taper and vibration characteristic parameters, the rail profile grinding parameters include the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding plan; S2: constructing an equivalent taper prediction model according to the rail line grade and the rail profile grinding parameters; S3: constructing a vibration characteristic parameter prediction model according to the rail line grade and the measured equivalent taper; S4: constructing a front-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade and the vibration characteristic parameters, and constructing a back-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade, the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point; S5: inputting the vibration characteristic parameters to be measured into the profile deviation value and polishing scheme prediction model to obtain a predicted profile deviation value and a predicted polishing scheme; S6: inputting the predicted profile deviation value and the predicted grinding plan into the equivalent taper prediction model to obtain a predicted equivalent taper; S7: inputting the predicted equivalent taper into the vibration characteristic parameter prediction model to obtain predicted vibration characteristic parameters; S8: cyclically executing S5 to S8 in sequence until the predicted vibration characteristic parameter meets the standard parameter value; S9: Grinding the target wheel rail according to the predicted vibration characteristic parameters that meet standard parameter values.

2. A high-speed railway rail profile grinding method according to claim 1, characterized in that: The S1 includes: The initial data set also includes a standard rail profile and a simulated rail profile, wherein: According to the standard rail profile and the simulated rail profile, obtaining the x-axis coordinate of the marking point and the deviation angle of the normal vector of the marking point; Obtaining the rail profile deviation value and the grinding scheme according to the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point; Obtaining the rail profile grinding amount according to the rail profile deviation value and the grinding scheme; Obtaining the left wheel rolling circle radius, the right wheel rolling circle radius and the wheel pair lateral displacement of the target wheel rail to obtain the measured equivalent taper; A vibration acceleration signal is acquired, and the vibration acceleration signal is converted into the vibration characteristic parameter according to a signal processing algorithm.

3. A high-speed railway rail profile grinding method according to claim 2, characterized in that: The S2 includes: Constructing a first training data set according to the rail line grade, the x-axis coordinate of the marking point, the normal vector deviation angle of the marking point, the rail profile deviation value, the rail profile grinding amount and the grinding scheme; Normalizing the first training data set to obtain a first standardized matrix; A first objective function for the equivalent taper prediction model is constructed, the first standardized matrix is ​​input into the equivalent taper prediction model, and the first objective function is iteratively optimized using a particle swarm algorithm, so as to finally obtain the optimized equivalent taper prediction model.

4. A high-speed railway rail profile grinding method according to claim 1, characterized in that: The S3 includes: Constructing a second training data set according to the rail line grade and the measured equivalent taper; Normalizing the second training data set to obtain a second standardized matrix; A second objective function for the vibration characteristic parameter prediction model is constructed, the second standardized matrix is ​​input into the vibration characteristic parameter prediction model, and the second objective function is iteratively optimized using a particle swarm algorithm, so as to finally obtain the optimized characteristic parameter prediction model.

5. The method for grinding the profile of a high-speed railway rail according to claim 1, characterized in that: The S4 includes: Constructing a third training data set according to the rail line grade and the vibration characteristic parameters; Normalizing the third training data set to obtain a third standardized matrix; Constructing a third objective function for the front-end network model, inputting the third standardized matrix into the front-end network model and iteratively optimizing the third objective function using a particle swarm algorithm, and finally obtaining an optimized front-end network model; Constructing a fourth training data set according to the measured equivalent taper, the x-axis coordinate of the marking point, and the normal vector deviation angle of the marking point; Normalizing the fourth training data set to obtain a fourth standardized matrix; Constructing a fourth objective function for the backend network model, inputting the fourth standardized matrix into the backend network model and iteratively optimizing the fourth objective function using a particle swarm algorithm, and finally obtaining an optimized backend network model; The optimized front-end network model and the optimized back-end network model are combined into the profile deviation value and polishing scheme prediction model.

6. A high-speed railway rail profile grinding method according to claim 1, characterized in that: The step S2 further includes: calculating the equivalent taper of the rail profile at different lateral displacements according to wheel-rail contact geometry theory and recording a critical value of the equivalent taper, wherein the critical value of the equivalent taper is used as one of the criteria for determining whether the grinding is qualified according to the predicted equivalent taper of the wheel-rail, specifically: At each marking point, i normal variations are set to obtain the normal deviation value between the simulated rail profile and the standard rail profile, wherein the normal deviation value is equal to the grinding amount of the simulated rail profile; The simulated rail profile grinding amounts are arranged and combined to obtain A combination of simulated rail profiles; Calculate the equivalent tapers λ1, λ2, ..., λ of the simulated rail profile under different lateral displacements for each combination. j And record the critical value λ of the equivalent taper min and λ max .

7. A high-speed railway rail profile grinding method according to claim 1, characterized in that: The S3 also includes: calculating the vibration acceleration signal of the rail profile under different lateral displacements through wheel dynamics simulation software and recording the maximum lateral vibration acceleration and the maximum vertical vibration acceleration as critical values ​​for judging the vibration characteristic parameters, wherein the critical value is used as one of the criteria for judging whether the grinding is qualified according to the predicted vibration characteristic parameters of the wheel and rail, specifically: The measured rail profile and the corresponding measured equivalent taper under different lateral displacements are imported into the dynamic simulation software to calculate the corresponding vibration acceleration signal; According to the vibration acceleration signal, record the corresponding maximum lateral vibration acceleration a xmax and the maximum vertical vibration acceleration a ymax .

8. A high-speed railway rail profile grinding system, characterized in that: The system comprises: The data acquisition module includes S1: acquiring an initial data set of a target wheel rail, wherein the initial data set includes rail line grade, rail profile grinding parameters, measured equivalent taper, and vibration characteristic parameters, wherein the rail profile grinding parameters include x-axis coordinates of a marking point, a normal vector deviation angle of a marking point, a rail profile deviation value, a rail profile grinding amount, and a grinding plan; The model building module includes S2: building an equivalent taper prediction model according to the rail line grade and the rail profile grinding parameters; S3: building a vibration characteristic parameter prediction model according to the rail line grade and the measured equivalent taper; S4: building a front-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade and the vibration characteristic parameters, and building a back-end network model of a profile deviation value and a grinding scheme prediction model according to the rail line grade, the measured equivalent taper, the x-axis coordinate of the marking point and the normal vector deviation angle of the marking point; The prediction module includes S5: inputting the vibration characteristic parameter to be measured into the profile deviation value and the grinding scheme prediction model to obtain the predicted profile deviation value and the predicted grinding scheme; S6: inputting the predicted profile deviation value and the predicted grinding scheme into the equivalent taper prediction model to obtain the predicted equivalent taper; S7: inputting the predicted equivalent taper into the vibration characteristic parameter prediction model to obtain the predicted vibration characteristic parameter; The judgment and grinding module includes S8: executing S5 to S8 in a loop in sequence until the predicted vibration characteristic parameters meet the standard parameter values; S9: grinding the target wheel rail according to the predicted vibration characteristic parameters that meet the standard parameter values.

9. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the executable instructions to implement a high-speed railway rail profile grinding method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute a high-speed railway rail profile grinding method as described in any one of claims 1 to 7.

Images