A method for optimizing turnout frog profile based on genetic optimization algorithm

Abstract

The present invention discloses a method for optimizing the profile of a turnout frog based on a genetic optimization algorithm, and belongs to the field of design optimization of the profile of a railway turnout frog area; the method comprises extracting sample cross-section data of a railway turnout frog area, parameterizing the sample cross-section curve by NURBS for characterization, encoding the parameters to establish an initial population, bringing the parameters into a dynamic model for optimization calculation, taking the wheel-rail vertical force when passing through the turnout as an objective function to find its optimal solution, determining the population individuals by establishing dynamic constraints and geometric constraints, and obtaining an optimized profile of the frog profile based on genetic optimization. The present invention can optimize the selection of the profile of the turnout frog area, thereby reducing the wheel-rail interaction force, thereby increasing the wheel-rail rolling fatigue release life, inhibiting the generation of cracks, and improving driving safety.

Description

Technical Field

[0001] The invention relates to the technical field of railway turnout frog area profile design optimization, in particular to a turnout frog profile optimization method based on a genetic optimization algorithm. Background Art

[0002] The subway is an important part of urban rail transit, and its safe, reliable and efficient operation is directly related to the safety of national property and people's lives. As an important part of the subway track system, the turnout area rails use non-uniform profiles. The change in the turnout rail profile leads to violent wheel-rail interaction when subway vehicles pass. Therefore, the service state of the turnout area rails is worse. The track disease in the turnout area has always been one of the key points and difficulties in the operation and maintenance of the subway.

[0003] The turnout profile is composed of curves of different curvatures, which are required to be smooth and continuous with local variation characteristics. The NURBS surface algorithm can flexibly control the geometric shape of the surface of an object, and has the characteristics of local support, strong convexity, standardization, and local modification. The NURBS surface algorithm is designed to describe the turnout profile, and the profile of the straight point rail of the turnout after wear is polished by modifying the weight factor of the control point.

[0004] Genetic Algorithm (GA) is a computational model of biological evolution that simulates the natural selection and genetic mechanism of Darwin's theory of biological evolution. It is a method of searching for the optimal solution by simulating the natural evolution process. Its main features are that it operates directly on structural objects, without the limitation of derivative and function continuity; it has inherent implicit parallelism and better global optimization ability; it adopts a probabilistic optimization method, which can automatically obtain and guide the optimized search space without the need for certain rules, and adjust the search direction adaptively. Summary of the invention

[0005] The main purpose of the present invention is to provide a method for optimizing the profile of a turnout frog based on a genetic optimization algorithm, with the goal of reducing the wheel-rail vertical force when passing through a turnout, collecting turnout cross-section sample data, characterizing the curved surface by NURBS parameterization, and then binary encoding the parameters, establishing a vehicle-turnout coupling dynamics model based on dynamics software, optimizing and calculating the wheel-rail vertical force based on a genetic optimization algorithm GA, determining the fitness of individual populations in turn, selecting, crossing, and mutating to obtain the optimal profile, thereby reducing the wheel-rail vertical force when passing through a turnout, reducing the plastic flow on the turnout surface, and inhibiting the generation of cracks.

[0006] The object of the present invention is achieved by the following technical solution: A frog profile optimization method based on a genetic optimization algorithm comprises the following steps:

[0007] Step 1: Count the key profiles of the frog, select multiple groups of measured wing rail and center rail profile data of the frog area, import them into the vehicle-turnout coupling dynamics model as initial input, and use them as the initial solution for optimization;

[0008] Step 2: parametrically characterize the turnout profile surface through NURBS surface fitting, generate value points, control points, and interpolation points, and optimize the profile design of the turnout straight point rail by modifying the weight factors of the control points;

[0009] Step 3: Establish a vehicle-turnout coupling dynamics model in multi-body dynamics software, and perform dynamic calculations to obtain an initial optimization solution;

[0010] Step 4: Use genetic algorithm to perform iterative optimization of the profile. After selecting, crossing over, and mutating the variable set in the current solution, a new solution is obtained. Then the objective function and constraint function are calculated again and compared with the global optimal solution. If the current solution is better, the global optimal solution is updated and then re-entered for iterative calculation. When the number of iterations reaches the set value, the optimal solution is output, that is, the optimized design profile.

[0011] As a further improvement of the present invention, in step 1, the tip of the frog core is taken as the zero point, and multiple cross sections at a certain distance backward are taken as the key cross sections of the frog core, and the frog profile within a certain position interval is formed by fitting and controlling the multiple key cross sections; the multiple key cross sections are typical representative profiles when the wheel load is transferred from the wing rail to the heart rail when the vehicle passes through the frog area, and actual cracks are also generated in this position interval.

[0012] As a further improvement of the present invention, the step 2 is specifically as follows:

[0013] The model value points are original data values, interpolation generates interpolation points, and then the control points are generated by NURBS algorithm to control them, which is a one-to-one correspondence;

[0014] A multi-time non-uniform rational B-spline curve is defined by n+1 polygonal control vertices, resulting in a piecewise rational polynomial vector function of the following form:

[0015]

[0016] Where: N i,p (u) is a rational basis function, which is represented by the node vector U = [u0,u1,…,u n+k+1 ], where u∈[0,1], di(i=0,1…,n) is the control point, ω i is the corresponding weight factor sequence;

[0017] According to the recursive formula defined by Cox and DeBoer, the e-order canonical B-spline basis function is obtained as:

[0018]

[0019] Given n interpolation points Q i , the control points are back-calculated and designed by the formula:

[0020]

[0021] Where: n+1 control points P i is an unknown quantity; Q k is the interpolation point; Q k Parameter value of It is obtained by chord length parameterization method, where l is the total chord length, And the node vector U is obtained by the formula:

[0022]

[0023] Set the parameter value Substituting into formula (8), the coordinates of the control points are obtained by converting formula (8):

[0024]

[0025] The NURBS surface is specified to be p times in the u direction and q times in the v direction, and its bivariate piecewise rational vector-valued function is as follows:

[0026]

[0027] where {P i,j}Constructed a control grid in two directions, {ω i,j} is the weight factor, {N i,p(u)} is the non-rational B-spline basis function that defines the knot vector U, {N j,q(v)} are the non-rational B-spline basis functions that define V, and:

[0028]

[0029] Here r = n + p + 1, s = m + q + 1;

[0030] Introduce piecewise rational basis functions:

[0031]

[0032] Then formula (9) can be rewritten as:

[0033]

[0034] As a further improvement of the present invention, in step 3, a vehicle-turnout coupling dynamic model is established in the dynamics software, wherein, for the heart rail, its 200-1000mm cross-section is controlled by NURBS curve interpolation, and the rest is supplemented by measured data, and the wing rail is filled entirely with measured data.

[0035] As a further improvement of the present invention, in step 4, the genetic algorithm is used to perform the profile optimization iteration as follows:

[0036] The obtained control points are encoded in the underlying binary system, and new data are obtained by segmentation, crossover and mutation. New data of a single section is generated by crossover mutation. All sections form a three-dimensional turnout, i.e., a new population. The fitness of the new population is determined by the constraint function, the unfit population is eliminated, and the minimum value of the objective function is sought among the fit population.

[0037] As a further improvement of the present invention, in step 4, the wheel-rail vertical force when passing the turnout is used as the objective function. At the same time, in order to ensure that the fitting profile curve is smooth and has no singular points or wavy points, when describing the rail profile in the fork area by the NURBS curve, the coordinates of the value point, the monotonicity of the rail profile curve, the profile concavity and convexity, and the standard profile ordinate are added as geometric constraints for optimization.

[0038] As a further improvement of the present invention, the coordinate constraint of the profile point is as follows: according to the measured profiles at different sections of the turnout rail, the vertical coordinates of the profile profile point and the vertical coordinates of the profile profile point of the standard profile of the section are used as the boundary conditions of the optimized profile to ensure that the optimized profile is consistent with the original profile; that is, the vertical coordinates of the profile profile point after optimization are equal to the vertical coordinates of the profile profile point of the standard profile;

[0039] The monotonicity constraint of rail profile curve is as follows: the monotonicity constraint of cross-section profile curve is expressed by piecewise function to ensure the monotonicity of frog core profile;

[0040] The specific constraints of profile concavity and convexity are as follows: the concavity and convexity characteristics of the rail profile in the frog area are analyzed, and the overall profile curve is set as a convex curve. The second-order derivative of the curve equation of the optimized profile is required to be not less than 0 to ensure that the profile is always a convex curve.

[0041] In addition, in order to ensure the running safety of the train with designed profile, the constraint function of the derailment coefficient is established according to the Nada l derailment coefficient calculation formula.

[0042]

[0043] Where: f d is the derailment coefficient objective function; Q and P are the wheel-rail lateral force and wheel-rail vertical force respectively; α is the maximum wheel flange angle; μ is the friction coefficient at the wheel flange; Coefit is the fitness matching parameter.

[0044] Beneficial effects:

[0045] 1. A method for optimizing the profile of a turnout frog based on a genetic optimization algorithm disclosed in the present invention. Since the turnout is composed of curves with different curvatures, the curves are required to be smooth and continuous, and have local variation characteristics. Based on NUMBS curve fitting, the key geometric profile of the frog is parameterized;

[0046] 2. A method for optimizing the profile of a turnout frog based on a genetic optimization algorithm disclosed in the present invention optimizes the objective function based on the genetic optimization algorithm, establishes geometric constraints and dynamic constraints, and repeatedly optimizes and iterates to obtain the optimal profile;

[0047] 3. The present invention discloses a method for optimizing the profile of a turnout frog based on a genetic optimization algorithm. On the basis of achieving beneficial effects 1 and 2, the method can manually grind the surface profile of the frog to an optimized profile, thereby significantly reducing the wheel-rail impact when passing the turnout, inhibiting the generation of frog cracks, and improving driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flow chart of an embodiment of the present invention;

[0049] Figure 2 It is a front view of the vehicle-turnout coupling dynamics model in an embodiment of the present invention;

[0050] Figure 3 A side view of a vehicle-turnout coupling dynamics model in an embodiment of the present invention;

[0051] Figure 4 A top view of a vehicle-turnout coupling dynamics model in an embodiment of the present invention;

[0052] Figure 5 Schematic diagram of NURBS surface in an embodiment of the present invention; wherein (a) is a control grid diagram, and (b) is a surface diagram;

[0053] Figure 6 It is a key cross-sectional profile diagram of the frog area in an embodiment of the present invention;

[0054] Figure 7 A three-dimensional modeling diagram of a curved surface obtained by fitting a key cross section of a frog area in an embodiment of the present invention;

[0055] Figure 8 It is a schematic diagram of the type value points, control points and interpolation points of the key cross section after parametric characterization using the NUMBS surface algorithm in an embodiment of the present invention;

[0056] Fig. 9 A diagram showing the dynamic modeling of a turnout in an embodiment of the present invention;

[0057] Fig.10 A vehicle dynamics modeling diagram in an embodiment of the present invention;

[0058] Fig.11 A vehicle trailer dynamics modeling diagram in an embodiment of the present invention;

[0059] Fig.12 A schematic diagram of a genetic algorithm in an embodiment of the present invention using segmentation, crossover and mutation to obtain new data;

[0060] Fig.13 A schematic diagram of monotonicity constraint in an embodiment of the present invention;

[0061] Fig.14 Graph showing the genetic iteration convergence process in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0063] Example

[0064] like Figure 1 As shown in the figure, a flow chart of a method for optimizing the profile of a turnout frog based on a genetic optimization algorithm is shown in the figure. The specific implementation steps are as follows:

[0065] Step 1: Take the tip of the frog core as the zero point, and the sections at 200mm, 300mm, 400mm, 500mm, 600mm, 700mm, 800mm, 900mm, and 1000mm backward as the key sections of the frog core. Statistical analysis of the key frog profiles is performed through actual measurement, such as Figure 6 As shown, the three-dimensional fitting is Figure 7 As shown in FIG. 1 , five sets of measured wing rail and center rail profile data of multiple frog areas are selected as initial inputs to the vehicle-turnout coupling dynamics model, and are used as the initial solution for optimization; the vehicle-turnout coupling dynamics model is shown in FIG. Figure 2-Figure 4 shown.

[0066] Step 2: parametrically characterize the turnout profile surface through NURBS surface fitting, and generate model points, control points, and interpolation points, such as Figure 8 As shown in the figure, the profile of the straight point rail of the turnout after wear is designed by modifying the weight factor of the control point;

[0067] NURBS spline control points:

[0068] The model value point is the original data value, and the interpolation point is generated by interpolation. Then the control point is generated by the NURBS algorithm to control it, which is a one-to-one correspondence.

[0069] The detailed definition of NURBS is given below. A multi-time non-uniform rational B-spline curve is defined by n+1 polygonal control vertices, resulting in a piecewise rational polynomial vector function in the following form.

[0070]

[0071] Where: N i,p (u) is a rational basis function, which is represented by the node vector U = [u0,u1,…,u n+k+1 ], where u∈[0,1], di(i=0,1…,n) is the control point, ω i is the corresponding weight factor sequence, and the control vertex d i Related,ω0,ωi>0,otherω i ≥0, to avoid the curve from degenerating into a point due to the reduction of weight factor. According to the recursive formula defined by Cox and De Boer, the e-order canonical B-spline basis function is obtained as:

[0072]

[0073] According to the above definition, when P = 3, the above formula is the mathematical expression of a cubic NURBS curve. The corresponding control points of the same number can be calculated by interpolation points, that is, given n interpolation points Q i , and the control points are back-calculated and designed through the formula.

[0074]

[0075] Where: n+1 control points P i is an unknown quantity; Q k is the interpolation point; Q k The parameter value is It is obtained by chord length parameterization method, where l is the total chord length, and And the node vector U is obtained by the formula.

[0076]

[0077] Set the parameter value Substituting into formula (8), the coordinates of the control points can be obtained by converting formula (8):

[0078]

[0079] This is the detailed definition of NUMBS surface

[0080] The NURBS surface is specified to be p times in the u direction and q times in the v direction, and its bivariate piecewise rational vector-valued function is as follows:

[0081]

[0082] where {P i,j}Constructed a control grid in two directions, {ω i,j} is the weight factor, {N i,p(u)} is the non-rational B-spline basis function that defines the knot vector U, {N j,q(v)} are the non-rational B-spline basis functions that define V, and:

[0083]

[0084] Here r=n+p+1, s=m+q+1.

[0085] Introduce piecewise rational basis functions:

[0086]

[0087] Then formula (9) can be rewritten as:

[0088]

[0089] like Figure 5 The figure shows a schematic diagram of a NURBS surface.

[0090] Step 3: Establish a vehicle-turnout coupling dynamics model in the multi-body dynamics software, and perform dynamic calculations to obtain the initial optimization solution; the established EMU trailer model is as follows: Fig.10 and Fig.11 As shown in Fig. 9 As shown, the cross-section of the turnout heart rail of 200-1000mm is controlled by NURBS curve interpolation, and the rest is supplemented by measured data. The wing rails are all filled with measured data, with a total of 75 wing rail sections and 64 heart rail sections.

[0091] Step 4: If Fig.12 As shown, a genetic algorithm is used to iterate the optimized profile, and the obtained control points are encoded in the underlying binary format. New data are obtained by segmentation, crossover and mutation. As shown in the figure, crossover mutation is used to generate new data for a single section. All sections form a three-dimensional turnout, i.e., a new population. The fitness of the new population is determined by a constraint function, and the unfit population is eliminated. The minimum value of the objective function is sought in the fit population.

[0092] The objective function is set as the wheel-rail vertical force when passing the turnout. At the same time, in order to ensure that the fitting profile curve is smooth and has no singular points or wavy points, the curve monotonicity, concavity and convexity and the standard profile ordinate are added as the geometric constraints for optimization when describing the rail profile in the frog area through the NURBS curve. In addition, in order to ensure the safety of train operation with the designed profile, the constraint function of the derailment coefficient is established according to the Nada l derailment coefficient calculation formula.

[0093] The rolling contact between the wheel and rail causes plastic flow of the rail surface material. When the plastic deformation exceeds the fracture strain of the rail surface material, cracks will form on the rail surface. With the continuous rolling contact load, the surface cracks can gradually extend downward to a depth of several millimeters. Related studies have shown that the magnitude of the wheel-rail impact force is one of the main factors affecting rolling contact fatigue. Therefore, the wheel-rail vertical force when passing the turnout is used as the objective function, with the aim of reducing the wheel-rail interaction force, thereby increasing the wheel-rail rolling fatigue relief life and inhibiting the occurrence of cracks.

[0094] The target requirement for the entire turnout composed of optimized single sections in dynamic calculation is to reduce the wheel-rail vertical force. The wheel-rail vertical force is Figure 2-Figure 4 The force magnitudes are shown in ZW1-ZW4.

[0095] Figure 2-Figure 4 In, M c 、M t 、M w are the mass of the vehicle body, frame and wheelset (kg); I wx ,I wy ,I wz They are the wheelset rolling, rotation and swaying inertia (kg·m 2 );I tx ,I ty ,I tz They are the frame rolling, nodding and shaking moments of inertia (kg·m 2 );I cx ,I cy ,I cz They are the vehicle body rolling, nodding and shaking motion inertia (kg·m 2 );K sx , K sy , K sz are the longitudinal, lateral and vertical stiffness of the secondary suspension on one side of the bogie (N / m); C sx , C sy , C sz are the longitudinal, lateral and vertical damping of the secondary suspension on one side of the bogie (N·s / m); K px , K py , K pzare the longitudinal, transverse and vertical stiffness of each axle box suspension (N / m); C px , C py , C pz They are the longitudinal, transverse and vertical damping of each axle box suspension (N·s / m).

[0096] 4.1 Type value point coordinate constraints:

[0097] According to the measured profiles of the turnout rail at different sections, the ordinates of the profile value points and the ordinates of the profile value points of the standard profile of the section are used as boundary conditions for optimizing the profile to ensure that the optimized profile is basically consistent with the original profile; (i.e. the ordinates of the profile value points after optimization are equal to the ordinates of the profile value points of the standard profile).

[0098] 4.2 Monotonicity constraints of rail profile curve:

[0099] Since the rail profile in the frog area is not symmetrical, the monotonicity constraint of the cross-sectional profile curve is expressed by a piecewise function to ensure the monotonicity of the frog core surface, such as Fig.13 As shown, the outline on the left side of the dotted line is monotonically increasing, and the outline on the right side of the dotted line is monotonically decreasing; (i.e., ensure that the left and right sides are monotonous respectively).

[0100] 4.3 Constraints on the Convexity of the Profile:

[0101] The concave-convex characteristics of the rail profile in the frog area are analyzed, and the overall profile curve is set as a convex curve. Therefore, the second-order derivative of the curve equation of the optimized profile is required to be no less than 0 to ensure that the profile is always a convex curve; (the curve equation is the spline curve equation obtained by the NUMBS method to ensure that the second-order derivative of the curve is no less than 0).

[0102] 4.4 Derailment coefficient constraints:

[0103] In order to ensure the safety of train operation with the designed profile, it is necessary to ensure that the derailment coefficient does not exceed the limit. That is, the derailment coefficient in the dynamic calculation result of the optimized profile cannot exceed the limit of 0.8, preferably not more than 0.6. In actual calculation, if the derailment coefficient exceeds 0.6, its fitness is reduced to 10%, and if it exceeds 0.8, the fitness is 0.

[0104]

[0105] Where: f d is the derailment coefficient objective function; Q and P are the wheel-rail lateral force and wheel-rail vertical force respectively; α is the maximum wheel flange angle; μ is the friction coefficient at the wheel flange; Coefit is the fitness matching parameter. (Previously, it was the geometric restriction of each cross-sectional spline curve of the optimized profile, and this is the overall restriction of the turnout, that is, the entire turnout composed of each cross-sectional spline curve in three-dimensional space after optimization meets the most basic requirements of dynamics, and the train does not derail).

[0106] Under the above four constraints, the objective function is iterated to select, cross, mutate and other operations on the variable set in the current solution to obtain a new solution. Then the objective function and constraint function are calculated again and compared with the global optimal solution. If the current solution is better, the global optimal solution is updated and then re-entered for iterative calculation. When the number of iterations reaches the set value, the optimal solution is output, that is, the optimized design profile. The iterative process is as follows: Fig.13 As shown in the figure, the maximum wheel-rail vertical force is reduced from 136.65kN to 115.35kN, and the optimization effect is significant.

[0107] The magnitude of the wheel-rail impact force is one of the main factors affecting rolling contact fatigue. This embodiment uses the wheel-rail vertical force when passing the turnout as the target function, aiming to reduce the wheel-rail interaction force, thereby increasing the wheel-rail rolling fatigue relief life and inhibiting the generation of cracks.

[0108] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope 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.

Claims

1. A frog profile optimization method based on genetic optimization algorithm, characterized in that: The following steps are involved: Step 1: Count the key profiles of the frog, select multiple groups of measured wing rail and center rail profile data of the frog area, import them into the vehicle-turnout coupling dynamics model as initial input, and use them as the initial solution for optimization; Step 2: parametrically characterize the turnout profile surface through NURBS surface fitting, generate value points, control points, and interpolation points, and optimize the profile design of the turnout straight point rail by modifying the weight factors of the control points; The step 2 is specifically as follows: The model value points are original data values, interpolation generates interpolation points, and then the control points are generated by NURBS algorithm to control them, which is a one-to-one correspondence; A multi-time non-uniform rational B-spline curve is defined by n+1 polygonal control vertices, resulting in a piecewise rational polynomial vector function of the following form: Where: N i,p (u) is a rational basis function, which is represented by the node vector U = [u0,u1,…,u n+k+1 ], where u∈[0,1], d i is a control point, where i = 1, 2, ..., n, ω i is the corresponding weight factor sequence; According to the recursive formula defined by Cox and DeBoer, the e-order canonical B-spline basis function is obtained as: Given n interpolation points Q i , the control points are back-calculated and designed by the formula: Where: n+1 control points P i is an unknown quantity; Q k is the interpolation point; Q k Parameter value of It is obtained by chord length parameterization method, where l is the total chord length, And the node vector U is obtained by the formula: Set the parameter value Substituting into formula (8), the coordinates of the control points are obtained by converting formula (8): The NURBS surface is specified to be p times in the u direction and q times in the v direction, and its two-variable piecewise rational vector-valued function is as follows: where {P i,j }Constructed a control grid in two directions, {ω i,j } is the weight factor, {N i,p(u) } is the non-rational B-spline basis function that defines the knot vector U, {N j,q(v) } are the non-rational B-spline basis functions that define V, and: Here r = n + p + 1, s = m + q + 1; Introduce piecewise rational basis functions: Then formula (9) can be rewritten as: Step 3: Establish a vehicle-turnout coupling dynamics model in multi-body dynamics software, and perform dynamic calculations to obtain an initial optimization solution; Step 4: Use genetic algorithm to perform iterative optimization of the profile, select, cross, mutate and other operations on the variable set in the current solution to obtain a new solution, then calculate the objective function and constraint function again, and compare them with the global optimal solution. If the current solution is better, the global optimal solution is updated, and then the updated global optimal solution is re-input for iterative calculation. When the number of iteration steps reaches the set value, the optimal solution is output, that is, the optimized design profile; In step 4, the genetic algorithm is used to perform the profile optimization iteration as follows: The obtained control points are encoded in the underlying binary system, and new data are obtained by segmentation, crossover and mutation. New data of a single section is generated by crossover mutation. All sections form a three-dimensional turnout, i.e., a new population. The fitness of the new population is determined by the constraint function, and the unfit population is eliminated. The minimum value of the objective function is sought in the fit population. In step 4, the wheel-rail vertical force when passing the turnout is used as the objective function. At the same time, in order to ensure that the fitting profile curve is smooth and has no singular points or wavy points, when describing the rail profile in the frog area through the NURBS curve, the coordinates of the value point, the monotonicity of the rail profile curve, the profile concavity and convexity, and the standard profile ordinate are added as the geometric constraints for optimization; The specific constraints of the coordinates of the profile points are as follows: according to the measured profiles of the turnout rail at different sections, the vertical coordinates of the profile profile points and the vertical coordinates of the profile profile points of the standard profile of the section are used as the boundary conditions of the optimized profile to ensure that the optimized profile is consistent with the original profile; that is, the vertical coordinates of the profile profile points after optimization are equal to the vertical coordinates of the profile profile points of the standard profile; The monotonicity constraint of rail profile curve is as follows: the monotonicity constraint of cross-section profile curve is expressed by piecewise function to ensure the monotonicity of frog core profile; The specific constraints of profile concavity and convexity are as follows: the concavity and convexity characteristics of the rail profile in the frog area are analyzed, and the overall profile curve is set as a convex curve. The second-order derivative of the curve equation of the optimized profile is required to be not less than 0 to ensure that the profile is always a convex curve. In addition, in order to ensure the running safety of the train with the designed profile, the constraint function of the derailment coefficient is established according to the Nadal derailment coefficient calculation formula; Where: f d is the derailment coefficient objective function; Q and P are the wheel-rail lateral force and wheel-rail vertical force respectively; α is the maximum wheel flange angle; μ is the friction coefficient at the wheel flange; Coefit is the fitness matching parameter.

2. The frog profile optimization method based on genetic optimization algorithm according to claim 1, characterized in that: In step 1, the tip of the frog core is taken as the zero point, and multiple sections at a certain distance backward are taken as the key sections of the frog core, and the frog profile within a certain position interval is formed by fitting and controlling the multiple key sections; the multiple key sections are typical representative profiles when the wheel load is transferred from the wing rail to the heart rail when the vehicle passes through the frog area, and actual cracks also occur in this position interval.

3. The frog profile optimization method based on genetic optimization algorithm according to claim 1, characterized in that: In step 3, a vehicle-turnout coupling dynamic model is established in the dynamics software, in which, for the heart rail, its 200-1000mm section is controlled by NURBS curve interpolation, 0-200mm and part of the section after 1000mm are supplemented by measured data, and the wing rail is filled entirely with measured data.

Images