Optimization Method for Dimension Parameters and Processing Parameters and Axle Bridge Manufactured by This Method

Abstract

The present invention discloses an optimization method for dimensional parameters and machining parameters, with the optimization object being design parameters; constructing the objective function of the integrated optimization model for shaft bridge design and manufacturing and the constraint function of design parameters; optimizing the shaft bridge dimensional parameters and shaft bridge machining parameters through sequential quadratic programming according to Formula 1, i.e., the integrated optimization model for shaft bridge design and manufacturing: solving to obtain the optimal shaft bridge dimensional parameters and shaft bridge machining parameters. Through iterative calculation according to Formula 1, the present invention can calculate the optimized values of each design parameter from the relevant parameters of the initial existing shaft bridge, which not only meets the constraint threshold but also makes it the smallest, so that the finally obtained shaft bridge dimensional parameters and shaft bridge machining parameters can achieve the optimal cost on the premise of ensuring the performance target. The present invention also provides a shaft bridge manufactured using this optimization method.

Description

Technical Field

[0001] The present invention relates to the technical field of machining, and particularly to the structure and machining method optimization technology of an axle bridge for a vehicle bogie. Background Art

[0002] With the rapid development of urban rail transit, low-floor vehicles have become an important choice for urban transportation due to their economic, environmental, convenient, and comfortable advantages.

[0003] As an important component of the low-floor vehicle bogie, the axle bridge supports the weight of the entire vehicle body. The independent wheels connected to both ends of the axle bridge rotate independently under the drive of the motor and drive the vehicle to travel. In order to reduce the design and manufacturing cost of the axle bridge and ensure the performance and safety of vehicle operation, high requirements are imposed on the structure, machining performance, and assembly accuracy with the independent wheels of the axle bridge.

[0004] The existing axle bridge includes an axle body in terms of structure. Spring seats are respectively provided at the left and right ends of the axle body. The spring seats extend in the front-back direction. An axle head is provided upward on the spring seat facing the axle body. A large axle neck is provided on the axle head in the direction away from the axle body, and a small axle neck is coaxially provided on the large axle neck in the direction away from the axle head; a spring seat center hole is provided on the spring seat other than the axle head.

[0005] The structure of the axle bridge is complex and irregular in shape, and the machining difficulty is relatively large. For the forging and welding blank of the axle bridge, turning, finish milling, and grinding processes are used for machining the axle neck part of the axle bridge. Since the axle bridge is a crankshaft-like part with a large size and its center of gravity is not on the center line of the axle neck, when turning and grinding the axle necks on both sides of the axle bridge, the axle bridge needs to rotate for machining, so it is easy to generate a large centrifugal force, and it is necessary to overcome the problems of poor axle neck dimensions, surface quality, and geometric tolerance caused by the large eccentricity and poor rigidity of the axle bridge.

[0006] The inventor found some relevant patent documents through domestic patent literature retrieval, mainly including the following:

[0007] The patent with the publication number CN205022599U discloses an "axle bridge structure for a low-floor rail vehicle". This structure includes two symmetrically arranged axle head assemblies, which are connected by a horizontal beam. The axle head assemblies are integrally cast, the horizontal beam is formed by splicing and welding metal plates, and the axle head assemblies and the horizontal beam are fixedly connected by welding. A connection seat for connecting with the wheel and a spring mounting seat for installing a primary spring are provided in both axle head assemblies. This utility model adopts an integral casting machining method, which has a lower machining cost and can improve production efficiency. Moreover, this structure can reduce the difficulty of the casting process, reduce costs, and also enable the axle bridge to have sufficient fatigue strength and good processability.

[0008] The patent publication number CN107953143A discloses the invention of "A low-floor rail vehicle axle bridge processing counterweight block (original counterweight tooling)", which relates to a low-floor rail vehicle axle bridge processing counterweight block (original counterweight tooling), including a counterweight frame and a counterweight bracket, the counterweight frame includes two left and right support plates, the front and rear sides of the support plates are provided with connecting plates, the connecting plates are provided with a first connecting hole and a second connecting hole, the counterweight bracket includes two parallel cross bars, a longitudinal bar is provided between the middle of the front and rear cross bars, the left and right ends of the cross bars are provided with connecting blocks, and the top surface of the connecting blocks is provided with a third connecting hole and a fourth connecting hole. The invention avoids the deviation of the size and shape and position tolerance of the journal due to eccentric rotation and the rigidity of the axle bridge, and solves the most difficult problem to solve in axle bridge processing.

[0009] The patent with publication number CN104759848A discloses a "manufacturing method for axle bridge for low-floor vehicles". The invention relates to a manufacturing method for axle bridge for low-floor vehicles, which includes the following steps in sequence: material selection - blank making - forging and forming using a forging die - normalizing - rough machining - ultrasonic flaw detection - heat treatment - shot blasting - correction and semi-finishing - aging treatment - finishing - magnetic particle flaw detection - spraying. The axle bridge for low-floor vehicles manufactured using the manufacturing method of the present invention has a journal, axle head, a mounting seat and axle body as a single component as an integral whole. The axle bridge is integrally forged and has few defects, basically no defects such as pores, sand holes, shrinkage, etc., a high production yield, superior finished product performance, and good mechanical properties. Due to its series of advantages, the integral forged axle bridge is suitable for mass production and use.

[0010] The above patent documents are all focused on the structural form or assembly performance of the axle bridge, without considering the design and manufacturing cost of the axle bridge. Although they have brought about an improvement in service performance, they have increased the cost of design and processing.

[0011] The cost and service performance optimization in the design and manufacturing process of the axle bridge are all technical development directions worthy of research, improvement and protection. Summary of the invention

[0012] The purpose of the present invention is to provide a method for optimizing dimensional parameters and processing parameters, by establishing a dual aggregation objective function for integrated optimization of design and manufacturing, and a dual conservative aggregation constraint function in the life and service process (life), so as to ensure the service performance of the axle bridge while reducing its total design and manufacturing cost.

[0013] To achieve the above object, the present invention provides an optimization method for dimensional parameters and machining parameters for manufacturing a shaft bridge. The shaft bridge includes a shaft body, with spring seats provided at both left and right ends of the shaft body, and the spring seats extend in the front-rear direction; an axle head is provided upwardly opposite to the spring seat of the shaft body, a large axle neck is provided on the axle head in the direction away from the shaft body, a small axle neck is coaxially provided on the large axle neck in the direction away from the axle head, and the diameter of the small axle neck is smaller than that of the large axle neck; the length of the small axle neck is l1, and the length of the large axle neck is l2.

[0014] A spring seat center hole is provided on the spring seat other than the axle head; the machining of the shaft bridge is to machine a forging and welding blank of the shaft bridge into a finished shaft bridge.

[0015] The spring seat, axle head, large axle neck, and small axle neck form a set of unilateral structures, and the unilateral structures are symmetrically distributed at both left and right ends of the shaft body.

[0016] The optimization object is the design parameters; construct the objective function of the integrated optimization model for the design and manufacture of the shaft bridge and the constraint function of the design parameters.

[0017] (1) The design parameters include:

[0018] The dimensional parameters of the shaft bridge (x1, x2, x3, x4, x5, x6),

[0019] The machining parameters (x7, x8, x9) when turning the small axle neck x1,

[0020] and the machining parameters (x 10 , x 11 , x 12 ) when turning the large axle neck x2;

[0021] Each design parameter has upper and lower bound constraints.

[0022] Among them, x1 is the diameter of the small axle neck, x2 is the diameter of the large axle neck, x3 is the width of the spring seat, x4 is the thickness of the spring seat, x5 is the height of the shaft body, and x6 is the width of the shaft body; x1 to x6 are the dimensional parameters of the shaft bridge;

[0023] x7 is the cutting speed when turning the small axle neck,

[0024] x8 is the feed rate when turning the small axle neck;

[0025] x9 is the depth of cut when turning the small axle neck;

[0026] x 10 is the cutting speed when turning the large axle neck;

[0027] x 11 is the feed rate when turning the large axle neck;

[0028] x 12is the depth of cut in turning when machining the large journal;

[0029] x7 to x 12 are the machining parameters of the axle bridge;

[0030] (2) The objective function of the integrated optimization model for the design and manufacture of the axle bridge is:

[0031] The total cost C of the design and manufacture of the axle bridge 总 , including the cost C S in the design stage and the cost C Z in the manufacturing stage;

[0032] (3) The constraint functions include:

[0033] The fatigue life constraint g1(x1, x2, x3, x4, x5, x6),

[0034] The maximum deformation constraint g2(x1, x2, x3, x4, x5, x6),

[0035] The maximum stress constraint g3(x1, x2, x3, x4, x5, x6),

[0036] The maximum deformation constraint g4(x1, x2, x3, x4, x5, x6, x7, x8, x9) when turning the small journal,

[0037] The maximum stress constraint g5(x1, x2, x3, x4, x5, x6, x7, x8, x9) when turning the small journal,

[0038] The maximum deformation constraint g6(x1, x2, x3, x4, x5, x6, x 10 , x 11 , x 12 ) and

[0039] The maximum stress constraint g7(x1, x2, x3, x4, x5, x6, x 10 , x 11 , x 12 ) when turning the large journal;

[0040] The part inside the parentheses of each constraint function is the design parameter, and the part outside the parentheses of each constraint function is the function name of the constraint function;

[0041] Each constraint function is an implicit function of its design parameters;

[0042] According to Formula 1, that is, the integrated optimization model for the design and manufacture of the axle bridge, the size parameters and machining parameters of the axle bridge are optimized by sequential quadratic programming:

[0043] min: C 总 = C S + CZ

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051]

[0052] In Formula 1, to are the thresholds of the corresponding constraint functions determined by the enterprise;

[0053] Taking the existing dimensional parameters and machining parameters of the axle bridge as the initial values of each design parameter, perform simulation iterative calculations according to Formula 1, and screen out the different values of each design parameter in the simulation iterative calculations based on the condition that each constraint function satisfies the corresponding threshold to obtain C 总 , and the values of each design parameter corresponding to the smallest calculated C 总 value are the optimal values of each design parameter.

[0054] It also includes an optimization method for the total design and manufacturing cost C 总 of the axle bridge:

[0055] The calculation formula for the total design and manufacturing cost C 总 of the axle bridge is C 总 = C S + C Z ;

[0056] The cost C S in the design stage = C 材 + C 人 , C 人 is the labor cost in the design stage; C 材 is the material cost;

[0057] The material cost C 材 is calculated according to Formula 2, and Formula 2 is:

[0058]

[0059] where V is the volume of the forging and welding blank of the axle bridge, ρ is the material density of the forging and welding blank of the axle bridge, C单价 is the unit price of the material for the forging and welding blank of the shaft bridge, l1 is the axial length of the small journal, l2 is the axial length of the large journal; l3 is the length of the shaft body; l4 is the total width of the shaft bridge; l5 is the height of the shaft head;

[0060] The manufacturing stage cost C of the shaft bridge Z Calculated according to Formula Three, and Formula Three is:

[0061] C Z = 2×C 小 + 2×C 大 + C 其它 ;

[0062] In Formula Three, C 小 is the total processing cost of the small journal, C 大 is the total processing cost of the large journal, C 其它 is the processing cost of other shaft bridge sections;

[0063] The total processing cost C of the small journal 小 includes the direct turning cost C M , the tool change cost C R and the tool wear cost C D , and the corresponding calculation formula is as follows:

[0064]

[0065]

[0066]

[0067] C M is determined by the actual cutting amount and the cost coefficient K0 during the processing of the small journal;

[0068] K0 represents the sum of the labor cost and the management cost per unit time;

[0069] K t represents the tool cost coefficient for processing the small journal;

[0070] t e represents the tool change time,

[0071] T p represents the tool life,

[0072] l1 represents the length of the small journal,

[0073] C 其它 includes the processing cost of the shaft body and the boring processing cost at the spring seat,

[0074] The total processing cost C of the large journal 大 includes the direct turning cost Tool change cost and tool wear cost The calculation formula is as follows:

[0075]

[0076]

[0077]

[0078] It is determined by the actual cutting amount and the cost coefficient K0 when machining the large journal;

[0079] K0 represents the sum of the labor cost and the management cost per unit time;

[0080] K t * represents the tool cost coefficient for machining the large journal;

[0081] t e represents the tool change time,

[0082] T p represents the tool life.

[0083] Construct a double aggregation model for the fatigue life of the axle bridge;

[0084] In Formula 1, the fatigue life constraint g1(x1,x2,x3,x4,x5,x6) is an implicit function of x1 to x6, and the specific value of g1 is obtained through experiments or finite element simulations;

[0085] Adopt a double aggregation model to approximately fit the implicit relationship function between the design parameters - fatigue life. The specific expression of the double aggregation model is Formula 4:

[0086]

[0087] In Formula 4, is the aggregated Kriging model, is the aggregated RBF model, is the aggregated SVR model; w K is the weight value of the aggregated Kriging model, w R is the weight value of the aggregated RBF model, w S is the weight value of the aggregated SVR model;

[0088] The construction of Formula 4 includes two steps. The first step is to construct and

[0089] (4.1) The expression of the aggregated Kriging model is specifically as follows:

[0090]

[0091] Among them, is the Kriging model with Corrgauss as the correlation model,

[0092] is the Kriging model with corrlin as the correlation model,

[0093] is the Kriging model with Cubic as the correlation model;

[0094] The calculation formulas for w1, w2, and w3 in the aggregated Kriging model are:

[0095]

[0096] In the calculation formulas for w1, w2, and w3, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; and are collectively referred to as correlation coefficient 's calculation formula is as follows:

[0097]

[0098] In the calculation formula of the correlation coefficient , n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft-bridge performance response at the i-th sample point, is the mean value of the true values of the performance responses at n sample points, is the predicted value of the shaft-bridge fatigue life at the i-th sample point, obtained by calling the j-th Kriging approximate model;

[0099] (4.2) The expression of the aggregated RBF model is specifically as follows:

[0100]

[0101] In the expression of the aggregated RBF model,

[0102] is the RBF model with Corrgauss as the kernel function,

[0103] It is an RBF model with linear as the kernel function,

[0104] It is an RBF model with Polynomial Kernel as the kernel function;

[0105] The calculation formulas for w4, w5, and w6 are as follows:

[0106]

[0107] In the calculation formulas for w4, w5, and w6, is the correlation coefficient of, is the correlation coefficient of, is the correlation coefficient of; The correlation coefficient The calculation formula is as follows:

[0108]

[0109] The correlation coefficient In the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft-bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life values at n sample points, is the predicted value of the shaft-bridge fatigue life at the i-th sample point, obtained by calling the j-th RBF approximate model;

[0110] (4.3) The expression of the aggregated SVR model is specifically:

[0111]

[0112] In the expression of the aggregated SVR model, is an SVR model with linear as the kernel function, is an SVR model with Polynomial Kernel as the kernel function, is an SVR model with RBF as the kernel function;

[0113] In the expression of the aggregated SVR model, the calculation formulas for w7, w8, and w9 are as follows:

[0114]

[0115] In the calculation formulas for w7, w8, and w9, is the correlation coefficient of, is the correlation coefficient of, is correlation coefficient; The correlation coefficient is calculated as follows:

[0116]

[0117] The correlation coefficient in the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life values at n sample points, is the predicted value of the axle bridge fatigue life at the i-th sample point, obtained by calling the j-th SVR approximate model;

[0118] The second step in constructing Equation 4 is to construct a double aggregation model to obtain Equation 4;

[0119] In Equation 4, w K , w S , w R is calculated as follows:

[0120]

[0121] w K , w S , w R in the calculation formula, is correlation coefficient, is correlation coefficient, is correlation coefficient; Let and be collectively referred to as R 2 , and here the correlation coefficient R 2 is calculated as follows:

[0122]

[0123] Here, in the calculation formula of the correlation coefficient R 2 , n is the number of sample points for constructing the approximate model, and y i is the true value of the axle bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life values at n sample points, is the predicted value of the axle bridge fatigue life at the i-th sample point, obtained by calling the j-th aggregation model.

[0124] Construct a double Kriging aggregation model for the maximum deformation and maximum stress in the axle bridge design and manufacturing stage The specific expression is Formula Five, and the implicit relationship between the shaft bridge size parameters, shaft bridge processing parameters and performance response is fitted through Formula Five;

[0125] Formula Five is:

[0126] In Formula Five, is the aggregated 0th-order Kriging model, is the aggregated 1st-order Kriging model, is the aggregated 2nd-order Kriging model; w 0K is 's weight value, w 1K is 's weight value, w 2K is 's weight value;

[0127] The construction of Formula Five includes two steps. The first step is to construct and

[0128] (5.1) The expression of the aggregated 0th-order Kriging model is:

[0129]

[0130] Among them, is the 0th-order Kriging model with Corrgauss as the correlation model,

[0131] is the 0th-order Kriging model with corrlin as the correlation model,

[0132] is the 0th-order Kriging model with Cubic as the correlation model;

[0133] In the aggregated 0th-order Kriging model, w 01 , w 02 , w 03 's calculation formula is:

[0134]

[0135] w 01 , w 02 , w 03 In the calculation formula of, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; The correlation coefficient 's calculation formula is as follows:

[0136]

[0137] Correlation coefficient In the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle bridge fatigue life at the i-th sample point (obtained through finite element simulation), is the mean value of the true fatigue life values at n sample points, is the predicted value of the axle bridge performance response at the i-th sample point, obtained by calling the j-th zero-order Kriging approximation model;

[0138] (5.2) The expression of the aggregated first-order Kriging model is specifically:

[0139]

[0140] In the expression of the aggregated first-order Kriging model,

[0141] is the first-order Kriging model with Corrgauss as the correlation model,

[0142] is the first-order Kriging model with corrlin as the correlation model,

[0143] is the first-order Kriging model with Cubic as the correlation model;

[0144] w 14 , w 15 , w 16 The calculation formula is as follows:

[0145]

[0146] w 14 , w 15 , w 16 In the calculation formula of, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; here the correlation coefficient The calculation formula is as follows:

[0147]

[0148] Correlation coefficient In the calculation formula, n is the number of sample points for constructing the approximate model, and yi is the true value of the shaft bridge performance response at the i-th sample point, is the predicted value of the shaft bridge performance response at the i-th sample point, obtained by calling the j-th first-order Kriging approximation model.

[0149] (5.3) The expression of the aggregated second-order Kriging model is specifically:

[0150]

[0151] In the expression of the aggregated second-order Kriging model, is the second-order Kriging model with Corrgauss as the correlation model, is the second-order Kriging model with corrlin as the correlation model, is the second-order Kriging model with Cubic as the correlation model;

[0152] w 27 , w 28 , w 29 The calculation formula is as follows:

[0153]

[0154] w 27 , w 28 , w 29 In the calculation formula of, is 's correlation coefficient; The correlation coefficient The calculation formula is as follows:

[0155]

[0156] The correlation coefficient In the calculation formula of, n is the number of sample points for constructing the approximation model, y i is the true value of the shaft bridge performance response at the i-th sample point, is the mean of the true values of the performance responses at n sample points, is the predicted value of the shaft bridge performance response at the i-th sample point, obtained by calling the j-th second-order Kriging approximation model;

[0157] The second step in constructing Formula Five is to construct a double-aggregated model based on the 0th-order Kriging, 1st-order Kriging, and 2nd-order Kriging aggregated models to obtain Formula Five;

[0158] In Formula Five, w 0K , w 1K , w 2K The calculation formula is as follows:

[0159]

[0160] w 0K , w 1K , w 2K In the calculation formula of, w 0K is the correlation coefficient of, w 1K is the correlation coefficient of, w 2K is the correlation coefficient of; Here, the correlation coefficient The calculation formula is as follows:

[0161]

[0162] Here, the correlation coefficient In the calculation formula of, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life at n sample points, is the predicted value of the shaft bridge performance response at the i-th sample point, obtained by calling the j-th aggregation model.

[0163] To construct the mathematical models of each constraint function and select sample data for simulation experiments, the specific steps are as follows:

[0164] First step, uniformly generate 10 6 candidate sample points in the entire design domain through grid sampling;

[0165] Second step, replace the sample point closest to x0 in the design domain with the given initial design point x0 as the first training sample point;

[0166] Third step, calculate the distances between the candidate sample points except x0 and the first training sample point x0, and select the sample point with the largest distance as the 2nd training sample point, designated as x1;

[0167] Fourth step, calculate the distances between the candidate sample points except x0 and x1 and x0 and the distance to x1 Select the sample point with the largest value as the 3rd training sample point, designated as x2;

[0168] Fifth step, calculate the distances between each candidate sample point and the training sample points x0, x1,..., x k-1 respectively Select the candidate sample point with the largest value as the (k + 1)-th training sample point;

[0169] Step 6: Repeat step 5. When the number of training sample points reaches 10n, sampling ends.

[0170] The present invention also discloses an axle bridge manufactured by the optimization method of the above-mentioned size parameters and processing parameters, wherein two large journals and two small journals at the left and right ends of the axle body are coaxially arranged, and the axis is higher than the upper surface of the axle body.

[0171] The various parts of the axle bridge are integrally cast structures. In each single-side structure, the centroids of the large journal and the small journal are located on the same horizontal line as the centroid of the shaft head.

[0172] The vertical section of the shaft head extending along the front-rear direction is a trapezoid that is small at the top and large at the bottom.

[0173] The bottom surface of the spring seat, the bottom surface of the shaft head and the bottom surface of the shaft body are located in the same plane; the left and right widths of the spring seat are the same as the left and right widths of the shaft head.

[0174] The present invention has the following advantages:

[0175] The present invention uses formula 1 to perform iterative calculations, and can calculate the optimal constraint threshold ( to ) of the axle bridge size parameters and axle bridge processing parameters (obtain the optimized values of each design parameter), and from the results obtained by previous iterative calculations, the results that satisfy the requirements of each inequality in formula 1 and make C 总 The final optimized values of the smallest axle bridge related parameters (axle bridge size parameters and axle bridge processing parameters) enable the final axle bridge size parameters and axle bridge processing parameters to achieve optimal cost while ensuring performance targets.

[0176] The present invention adopts the dual aggregation model of formula 4 to approximate the implicit relationship function of design parameters-fatigue life, which solves the problem of excessively high calculation and experimental costs caused by directly calling the fatigue life response value at the iterative design point, and significantly reduces the cost of optimization design.

[0177] The present invention adopts the dual Kriging aggregation model of Formula 5 to approximate the implicit relationship function of design parameters-maximum deformation and design parameters-maximum stress in the design and manufacturing stage of the axle bridge, which solves the problem of excessive calculation and experimental costs caused by directly calling the fatigue life response value at the iterative design point, and significantly reduces the cost of optimization design.

[0178] The axes of the two large journals and the two small journals are higher than the upper surface of the axle body, so that the entire axle bridge structure is U-shaped, which can reduce the distance between the floor surface and the rail surface of the floor vehicle and is suitable for use in low-floor vehicles.

[0179] The centroids of the large journal and the small journal are both on the same horizontal line as the centroid of the shaft head, so that the unilateral structure and the shaft bridge as a whole have good mechanical properties. Brief Description of the Drawings

[0180] Figure 1 is a schematic diagram of the shaft bridge structure in the present invention;

[0181] Figure 2 is the overall optimization flow chart of the present invention;

[0182] Figure 3 is a schematic diagram for constructing a double aggregation model of the fatigue life of the shaft bridge;

[0183] Figure 4 is a flow chart for constructing a double Kriging aggregation model of the maximum deformation and maximum stress in the shaft bridge design and manufacturing stage;

[0184] Figure 5 is a schematic diagram of the sampling process. Detailed Description of the Invention

[0185] As Figures 1 to 5 shown, the present invention discloses an optimization method for dimensional parameters and machining parameters. The shaft bridge includes a shaft body 1. Spring seats 2 are respectively provided at the left and right ends of the shaft body 1, and the spring seats 2 extend in the front-rear direction; an axle head 3 is provided upwardly opposite to the spring seat 2 of the shaft body 1. A large journal 4 is provided on the axle head 3 in a direction away from the shaft body 1, and a small journal 5 is coaxially provided on the large journal 4 in a direction away from the axle head 3. The diameter of the small journal 5 is smaller than that of the large journal 4; the length of the small journal 5 is l1, and the length of the large journal 4 is l2. The units of both l1 and l2 are millimeters.

[0186] A spring seat center hole 6 is provided on the spring seat 2 other than the axle head 3; the machining of the shaft bridge is to machine the forging and welding blank of the shaft bridge into a finished shaft bridge;

[0187] The spring seat 2, the axle head 3, the large journal 4 and the small journal 5 form a set of unilateral structures, and the unilateral structures are symmetrically distributed at the left and right ends of the shaft body 1;

[0188] The optimization object is the design parameters; the objective function of the integrated optimization model for the design and manufacturing of the shaft bridge and the constraint function of the design parameters are constructed;

[0189] (1) The design parameters include:

[0190] The dimensional parameters (x1, x2, x3, x4, x5, x6) of the shaft bridge,

[0191] The machining parameters (x7, x8, x9) when turning the small journal 5x1 and

[0192] The machining parameters (x 10 , x 11,x 12 )

[0193] Each design parameter has upper and lower bound constraints;

[0194] Among them, x1 is the diameter of the small journal 5, x2 is the diameter of the large journal 4, x3 is the width of the spring seat 2, x4 is the thickness of the spring seat 2, x5 is the height of the shaft body 1, and x6 is the width of the shaft body 1; the units of x1 to x6 are all millimeters (i.e., mm); x1 to x6 are the shaft bridge dimension parameters;

[0195] x7 is the cutting speed for turning when machining the small journal 5;

[0196] x8 is the feed rate for turning when machining the small journal 5;

[0197] x9 is the depth of cut for turning when machining the small journal 5;

[0198] x 10 is the cutting speed for turning when machining the large journal 4;

[0199] x 11 is the feed rate for turning when machining the large journal 4;

[0200] x 12 is the depth of cut for turning when machining the large journal 4.

[0201] x7 to x 12 are the shaft bridge machining parameters;

[0202] x7 and x 10 are both in meters per minute (i.e., m / min);

[0203] x8 and x 11 are both in millimeters per revolution (i.e., mm / r);

[0204] x9 and x 12 are both in millimeters (i.e., mm);

[0205] (2) The objective function of the integrated optimization model for shaft bridge design and manufacturing is:

[0206] The total cost C of shaft bridge design and manufacturing 总 , including the cost C S in the design stage and the cost C Z in the manufacturing stage;

[0207] C 总 、C S and C Z are all in yuan;

[0208] (3) The constraint functions include:

[0209] Fatigue life constraint \(g_1(x_1,x_2,x_3,x_4,x_5,x_6)\),

[0210] Maximum deformation constraint \(g_2(x_1,x_2,x_3,x_4,x_5,x_6)\),

[0211] Maximum stress constraint \(g_3(x_1,x_2,x_3,x_4,x_5,x_6)\),

[0212] Maximum deformation constraint \(g_4(x_1,x_2,x_3,x_4,x_5,x_6,x_7,x_8,x_9)\) when turning the small journal 5,

[0213] Maximum stress constraint \(g_5(x_1,x_2,x_3,x_4,x_5,x_6,x_7,x_8,x_9)\) when turning the small journal 5,

[0214] Maximum deformation constraint \(g_6(x_1,x_2,x_3,x_4,x_5,x_6,x 10 ,x 11 ,x 12 ) and

[0215] Maximum stress constraint \(g_7(x_1,x_2,x_3,x_4,x_5,x_6,x 10 ,x 11 ,x 12 );

[0216] The part inside the parentheses of each constraint function is the design parameter, and the part outside the parentheses of each constraint function is the function name of the constraint function;

[0217] Each constraint function is an implicit function of its design parameters;

[0218] The size parameters or machining parameters of each axle bridge inside the parentheses of each constraint function are the parameter combinations of the constraint function;

[0219] According to Formula 1, that is, the integrated optimization model for axle bridge design and manufacturing, the size parameters and machining parameters of the axle bridge are optimized by sequential quadratic programming: the optimal size parameters and machining parameters of the axle bridge are obtained by solving.

[0220] min: C 总 =C S +C Z

[0221]

[0222]

[0223]

[0224]

[0225]

[0226]

[0227]

[0228]

[0229] In Formula 1, to are the thresholds of the corresponding constraint functions determined by the enterprise;

[0230] Taking the existing dimensional parameters and machining parameters of the axle bridge as the initial values of each design parameter, perform simulation iterative calculations according to Formula 1, and screen out the different values of each design parameter in the simulation iterative calculations with the condition that each constraint function satisfies the corresponding threshold to obtain C 总 , and the values of each design parameter corresponding to the minimum calculated C 总 value are the optimal values of each design parameter.

[0231] In Formula 1, there are 7 inequalities in total for g1 to g7. Among the 7 inequalities, the right side of the inequality sign (greater than sign or less than sign) is the threshold of the constraint function determined by the enterprise; the expression on the left side of the inequality sign is the constraint function, which is the constraint function value obtained by finite element simulation calculation of the threshold under the corresponding design parameter combination.

[0232] The fatigue life constraint g1(x1, x2, x3, x4, x5, x6) and are in years; represents the minimum service life constraint condition of the axle bridge required by the enterprise;

[0233] The maximum deformation constraint g2(x1, x2, x3, x4, x5, x6) and are in millimeters; represents the allowable deformation value constraint condition of the axle bridge required by the enterprise;

[0234] The maximum stress constraint g3(x1, x2, x3, x4, x5, x6) and are in megapascals; represents the allowable stress value constraint condition of the axle bridge required by the enterprise;

[0235] The maximum deformation constraint g4(x1, x2, x3, x4, x5, x6, x7, x8, x9) during turning of the small axle neck 5 and are in millimeters; represents the allowable deformation value constraint condition during machining of the small axle neck 5 of the axle bridge, and the optimal value according to machining experience is one-thousandth of the diameter of the small axle neck 5;

[0236] The maximum stress constraint g5 (x1, x2, x3, x4, x5, x6, x7, x8, x9) when turning the small journal 5 and The unit is MPa; It represents the allowable stress value constraint condition when machining the small journal 5 of the axle bridge. The preferred value is one third of the yield strength of the axle bridge material according to machining experience.

[0237] The maximum deformation constraint g6 (x1, x2, x3, x4, x5, x6, x 10 ,x 11 ,x 12 )and The unit is millimeter; It indicates the allowable deformation value constraint condition when machining the large journal 4 of the axle bridge required by the enterprise. The preferred value is one thousandth of the diameter of the large journal 4 according to machining experience.

[0238] The maximum stress constraint g7 (x1, x2, x3, x4, x5, x6, x7) when turning the large journal 4 10 ,x 11 ,x 12 )and The unit is MPa; It represents the allowable stress value constraint condition required by the enterprise when processing the large journal 4. According to processing experience, the preferred value is one-third of the yield strength of the shaft bridge material.

[0239] The present invention uses formula 1 to perform iterative calculations, and can calculate the constraint threshold ( to ) of the axle bridge size parameters and axle bridge processing parameters (obtain the optimized values of each design parameter), and from the results obtained by previous iterative calculations, the results that satisfy the inequality constraints in formula 1 and make C 总 The final optimized values of the smallest axle bridge related parameters (axle bridge size parameters and axle bridge processing parameters) enable the final axle bridge size parameters and axle bridge processing parameters to achieve optimal cost while ensuring performance targets.

[0240] It also includes the total cost of axle bridge design and manufacturing C 总 Optimization method:

[0241] Total cost of axle bridge design and manufacturing C 总 The calculation formula is C 总 =C S +C Z ;

[0242] Design stage cost S =C 材 +C人 , C 人 is the labor cost during the design stage; C 材 is the material cost;

[0243] The material cost C 材 is calculated according to Formula 2, and Formula 2 is:

[0244]

[0245] where V is the volume of the rough shaft-bridge forging and welding part, ρ is the material density of the rough shaft-bridge forging and welding part, C 单价 is the unit price of the material of the rough shaft-bridge forging and welding part (yuan / kg), l1 is the length of the small journal 5, l2 is the length of the large journal 4; l3 is the length of the shaft body 1; l4 is the total width of the shaft bridge (i.e., the maximum width of the shaft bridge); l5 is the height of the shaft head 3;

[0246] The cost C during the manufacturing stage of the shaft bridge Z is calculated according to Formula 3, and Formula 3 is:

[0247] C Z = 2×C 小 + 2×C 大 + C 其它 ;

[0248] In Formula 3, C 小 is the total processing cost of the small journal 5, C 大 is the total processing cost of the large journal 4, C 其它 is the processing cost of other shaft bridge sections;

[0249] The total processing cost C of the small journal 5 小 includes the direct turning cost C M (i.e., the processing cost during turning), the tool change cost C R and the tool wear cost C D , and the corresponding calculation formula is as follows:

[0250]

[0251]

[0252]

[0253] C M is determined by the actual cutting amount and the cost coefficient K0 during the processing of the small journal 5;

[0254] K0 represents the sum of the labor cost and the management cost per unit time, with the unit of yuan / minute, and depends on the actual value of the specific enterprise;

[0255] K tIt represents the tool cost coefficient for machining the small journal 5; the unit is yuan / cutting edge, which is determined by the designer according to the actual machining experience of the enterprise;

[0256] t e It represents the tool change time, and the unit is minute;

[0257] T p It represents the tool life, and the unit is minute;

[0258] l1 represents the length of the small journal 5, and the unit is millimeter;

[0259] C 其它 It includes the machining cost of the shaft body 1 and the boring cost (the center hole 6 of the spring seat) at the spring seat 2, which is independent of the machining parameters and is a constant. The specific value of C is determined through experiments. 其它 The specific value.

[0260] The total machining cost C of the large journal 4 大 It includes the direct turning cost The tool change cost and the tool wear cost Its calculation formula is as follows:

[0261]

[0262]

[0263]

[0264] It is determined by the actual cutting amount and the cost coefficient K0 during the machining of the large journal 4;

[0265] K0 represents the sum of the labor cost and the management cost per unit time, and the unit is yuan / minute, which depends on the actual value of the specific enterprise;

[0266] K t * It represents the tool cost coefficient for machining the large journal 4; the unit is yuan / cutting edge, which is determined by the designer according to the actual machining experience of the enterprise;

[0267] t e It represents the tool change time, and the unit is minute;

[0268] T p It represents the tool life, and the unit is minute.

[0269] Construct a double aggregation model for the fatigue life of the axle bridge;

[0270] In Formula 1, the fatigue life constraint g1(x1, x2, x3, x4, x5, x6) is an implicit function of x1 to x6, and the specific value of g1 is obtained through experiments or finite element simulations;

[0271] If the fatigue life response value at the iterative design point is directly called in the optimization, the calculation or experimental cost is too high and difficult to accept in practice. Therefore, the present invention adopts a dual aggregation model to approximately fit the implicit relationship function between the design parameters and the fatigue life. The dual aggregation model has the specific expression as Formula Four:

[0272]

[0273] In Formula Four, is the aggregated Kriging model, is the aggregated RBF model, is the aggregated SVR model; w K is the weight value of the aggregated Kriging model, w R is the weight value of the aggregated RBF model, w S is the weight value of the aggregated SVR model;

[0274] The construction of Formula Four includes two steps. The first step is to construct and

[0275] (4.1) The specific expression of the aggregated Kriging model is:

[0276]

[0277] Among them, is the Kriging model with Corrgauss as the correlation model,

[0278] is the Kriging model with corrlin as the correlation model,

[0279] is the Kriging model with Cubic as the correlation model;

[0280] The calculation formulas for w1, w2, and w3 in the aggregated Kriging model are:

[0281]

[0282] In the calculation formulas for w1, w2, and w3, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; Combine and collectively referred to as Correlation coefficient The calculation formula is as follows:

[0283]

[0284] Correlation coefficient In the calculation formula of the correlation coefficient, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle-bridge fatigue life at the i-th sample point (obtained through finite element simulation), is the mean value of the true fatigue life values at n sample points, is the predicted value of the axle-bridge fatigue life at the i-th sample point, obtained by calling the j-th Kriging approximate model;

[0285] (4.2) The expression of the aggregated RBF model is specifically:

[0286]

[0287] In the expression of the aggregated RBF model,

[0288] is an RBF model with Corrgauss as the kernel function,

[0289] is an RBF model with linear as the kernel function,

[0290] is an RBF model with Polynomial Kernel as the kernel function;

[0291] The calculation formulas for w4, w5, and w6 are as follows:

[0292]

[0293] In the calculation formulas for w4, w5, and w6, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; The correlation coefficient The calculation formula is as follows:

[0294]

[0295] Correlation coefficient In the calculation formula of the correlation coefficient, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle-bridge fatigue life at the i-th sample point (obtained through finite element simulation), is the mean of the true fatigue life values at n sample points, is the predicted value of the axle bridge fatigue life at the i-th sample point, obtained by calling the j-th RBF approximation model;

[0296] (4.3) The expression of the aggregated SVR model is specifically:

[0297]

[0298] In the expression of the aggregated SVR model, is an SVR model with the linear kernel function, is an SVR model with the Polynomial Kernel function, is an SVR model with the RBF kernel function;

[0299] In the expression of the aggregated SVR model, the calculation formulas for w7, w8, and w9 are as follows:

[0300]

[0301] In the calculation formulas for w7, w8, and w9, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; The correlation coefficient 's calculation formula is as follows:

[0302]

[0303] The correlation coefficient 's calculation formula, n is the number of sample points for constructing the approximation model, y i is the true value of the axle bridge fatigue life at the i-th sample point (obtained through finite element simulation), is the mean of the true fatigue life values at n sample points, is the predicted value of the axle bridge fatigue life at the i-th sample point, obtained by calling the j-th SVR approximation model;

[0304] The second step in constructing Formula 4 is to construct a double-aggregation model to obtain Formula 4;

[0305] In Formula 4, w K , w S , w R 's calculation formulas are as follows:

[0306]

[0307] wK , w S , w R In the calculation formula of is the correlation coefficient of is the correlation coefficient of is the correlation coefficient of; Let and be collectively referred to as Here, the correlation coefficient The calculation formula is as follows:

[0308]

[0309] Here, the correlation coefficient R 2 In the calculation formula of, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle bridge fatigue life at the i-th sample point (obtained through finite element simulation), is the mean value of the true fatigue life values at n sample points, is the predicted value of the axle bridge fatigue life at the i-th sample point, obtained by calling the j-th aggregation model.

[0310] The present invention uses the double aggregation model of Formula Four to approximately fit the implicit relationship function between design parameters - fatigue life, solving the problem of excessive calculation and experimental costs caused by directly calling the fatigue life response values at iterative design points, and significantly reducing the cost of optimization design.

[0311] Construct a double Kriging aggregation model for the maximum deformation and maximum stress in the design and manufacturing stage of the axle bridge The specific expression is Formula Five, and the implicit relationship between the axle bridge size parameters, axle bridge processing parameters, and performance responses (maximum stress, maximum deformation) is fitted through Formula Five;

[0312] Formula Five is:

[0313] In Formula Five, is the aggregated 0th-order Kriging model, is the aggregated 1st-order Kriging model, is the aggregated 2nd-order Kriging model; w 0K is the weight value of, w 1K is the weight value of, w 2K is the weight value of;

[0314] The construction of Formula Five includes two steps. The first step is to construct and

[0315] (5.1) The expression of the aggregated zero - order Kriging model is:

[0316]

[0317] where is the zero - order Kriging model with Corrgauss as the correlation model,

[0318] is the zero - order Kriging model with corrlin as the correlation model,

[0319] is the zero - order Kriging model with Cubic as the correlation model;

[0320] In the aggregated zero - order Kriging model, w 01 , w 02 , w 03 The calculation formula is:

[0321]

[0322] w 01 , w 02 , w 03 In the calculation formula of is the correlation coefficient of , is the correlation coefficient of , is the correlation coefficient of ; The correlation coefficient The calculation formula is as follows:

[0323]

[0324] In the calculation formula of the correlation coefficient , n is the number of sample points for constructing the approximate model, y i is the true value (obtained through finite - element simulation) of the shaft - bridge performance response (maximum stress, maximum deformation) at the i - th sample point, is the mean of the true values of the performance responses (stress and deformation) at n sample points, is the predicted value of the shaft - bridge performance response (stress and deformation) at the i - th sample point, obtained by calling the j - th zero - order Kriging approximate model;

[0325] (5.2) The specific expression of the aggregated first - order Kriging model is:

[0326]

[0327] In the expression of the aggregated first - order Kriging model,

[0328] is the first - order Kriging model with Corrgauss as the correlation model,

[0329] is the first - order Kriging model with corrlin as the correlation model,

[0330] is the first - order Kriging model with Cubic as the correlation model;

[0331] w 14 , w 15 , w 16 The calculation formulas of, w

[0332]

[0333] w 14 , w 15 , w 16 In the calculation formulas of, w is the correlation coefficient of, is the correlation coefficient of, is the correlation coefficient of; Here, the correlation coefficient The calculation formula is as follows:

[0334]

[0335] The correlation coefficient In the calculation formula of, n is the number of sample points for constructing the approximate model, y i is the true value (obtained through finite - element simulation) of the shaft - bridge performance response (stress and deformation) at the i - th sample point, is the predicted value of the shaft - bridge performance response (stress and deformation) at the i - th sample point, obtained by calling the j - th first - order Kriging approximate model.

[0336] (5.3) The expression of the aggregated second - order Kriging model is specifically:

[0337]

[0338] In the expression of the aggregated second - order Kriging model, is the second - order Kriging model with Corrgauss as the correlation model, is the second - order Kriging model with corrlin as the correlation model, It is a second-order Kriging model with Cubic as the relevant model;

[0339] w 27 ,w 28 ,w 29 The calculation formula of is as follows:

[0340]

[0341] w 27 ,w 28 ,w 29 In the calculation formula of, is The correlation coefficient; The correlation coefficient The calculation formula of is as follows:

[0342]

[0343] The correlation coefficient In the calculation formula of, n is the number of sample points for constructing the approximate model, and y i Is the true value (obtained through finite element simulation) of the axle-bridge performance response (maximum stress, maximum deformation) at the i-th sample point, Is the mean value of the true values of the performance responses (maximum stress, maximum deformation) at n sample points, Is the predicted value of the axle-bridge performance response (maximum stress, maximum deformation) at the i-th sample point, obtained by calling the j-th second-order Kriging approximate model;

[0344] The second step in constructing Formula Five is to construct a double aggregation model based on the aggregated model of the 0th-order Kriging, 1st-order Kriging, and 2nd-order Kriging To obtain Formula Five;

[0345] In Formula Five, w 0K ,w 1K ,w 2K The calculation formula of is as follows:

[0346]

[0347] w 0K ,w 1K ,w 2K In the calculation formula of, w 0K Is The correlation coefficient of, w 1K Is The correlation coefficient of, w 2K Is The correlation coefficient of; Here, the correlation coefficient The calculation formula of is as follows:

[0348]

[0349] In the calculation formula of the correlation coefficient here n is the number of sample points for constructing the approximate model, and y i is the true value (obtained through finite element simulation) of the shaft bridge performance response (maximum stress, maximum deformation) at the i-th sample point. is the mean value of the true values of the performance responses (maximum stress, maximum deformation) at n sample points. is the predicted value of the shaft bridge performance response (maximum stress, maximum deformation) at the i-th sample point, obtained by calling the j-th aggregated model.

[0350] To construct the mathematical models of the constraint functions and select sample data for simulation experiments, the following experimental design method is adopted in this patent, and its flowchart is as Figure 5 shown. The specific steps are as follows:

[0351] First step, 10 6 candidate sample points are uniformly generated in the entire design domain through grid sampling;

[0352] Second step, the given initial design point x0 is used to replace the sample point closest to x0 in the design domain as the first training sample point;

[0353] Third step, calculate the distances between the candidate sample points except x0 and the first training sample point x0, and select the sample point with the largest distance as the 2nd training sample point, designated as x1;

[0354] Fourth step, calculate the distances between the candidate sample points except x0 and x1 and x0 and the distance to x1 Select the sample point with the largest distance as the 3rd training sample point, designated as x2;

[0355] Fifth step, calculate the distances between each candidate sample point and the training sample points x0, x1,..., x k-1 respectively Select the candidate sample point with the largest distance as the (k + 1)-th training sample point;

[0356] Sixth step, repeat the fifth step. When the number of training sample points reaches 10n (n is the dimension of the design space), the sampling ends.

[0357] The present invention also discloses a shaft bridge manufactured by the optimization method using the above-mentioned dimensional parameters and processing parameters. The two large journal bearings 4 and the two small journal bearings 5 at the left and right ends of the shaft body 1 are coaxially arranged, and this axis is higher than the upper surface of the shaft body 1.

[0358] The axes of the two large journal bearings 4 and the two small journal bearings 5 are higher than the upper surface of the axle body 1, so that the whole axle bridge structure is U-shaped, which can reduce the distance between the floor surface of the floor vehicle and the rail surface and is suitable for use in low-floor vehicles.

[0359] The various parts of the axle bridge are of an integral forging and welding structure. In each single-sided structure, the centers of mass of the large journal bearing 4 and the small journal bearing 5 are on the same horizontal straight line as the center of mass of the axle head 3.

[0360] The centers of mass of the large journal bearing 4 and the small journal bearing 5 are on the same horizontal straight line as the center of mass of the axle head 3, so that the single-sided structure and the whole axle bridge have good mechanical properties.

[0361] The vertical section of the axle head 3 extending in the front-back direction is trapezoidal with a smaller upper part and a larger lower part. The front-back direction is the direction perpendicular to the axes of the large journal bearing 4 and the small journal bearing 5 in the horizontal plane.

[0362] The bottom surface of the spring seat 2, the bottom surface of the axle head 3 and the bottom surface of the axle body 1 are in the same plane; the left-right width of the spring seat 2 is the same as the left-right width of the axle head 3.

[0363] The above embodiments are only used to illustrate rather than limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the present invention can still be modified or equivalently replaced, and any modification or partial replacement without departing from the spirit and scope of the present invention shall be covered by the scope of the claims of the present invention.

Claims

1. Optimization method for dimensional parameters and machining parameters. The axle bridge includes an axle body. Spring seats are respectively provided at both left and right ends of the axle body, and the spring seats extend in the front-back direction. Opposite to the axle body, a shaft head is provided upward on the spring seat. A large shaft neck is provided on the shaft head in the direction away from the axle body. A small shaft neck is coaxially provided on the large shaft neck in the direction away from the shaft head, and the diameter of the small shaft neck is smaller than that of the large shaft neck. The length of the small shaft neck is l1, and the length of the large shaft neck is l2. A spring seat center hole is provided on the spring seat other than the shaft head. The machining of the axle bridge is to machine the forging and welding blank of the axle bridge into a finished axle bridge. The spring seat, shaft head, large shaft neck, and small shaft neck form a set of unilateral structures, and the unilateral structures are symmetrically distributed at both left and right ends of the axle body. It is characterized in that: The optimization object is the design parameters. The objective function of the integrated optimization model for the design and manufacture of the axle bridge and the constraint function of the design parameters are constructed. (1) The design parameters include: The dimensional parameters of the axle bridge (x1, x2, x3, x4, x5, x6), The machining parameters (x7, x8, x9) when turning the small shaft neck x1, and the machining parameters (x 10 , x 11 , x 12 ) when turning the large journal x2; Each design parameter has upper and lower bound constraints. Among them, x1 is the diameter of the small shaft neck, x2 is the diameter of the large shaft neck, x3 is the width of the spring seat, x4 is the thickness of the spring seat, x5 is the height of the axle body, and x6 is the width of the axle body. x1 to x6 are the dimensional parameters of the axle bridge. x7 is the cutting speed of turning when machining the small shaft neck, x8 is the feed rate of turning when machining the small shaft neck, x9 is the depth of cut of turning when machining the small shaft neck. x 10 is the cutting speed for turning during machining of the large journal; x 11 is the feed rate for turning when machining large journal x 12 is the depth of cut in turning when machining a large journal; x7 to x 12 are the axle bridge machining parameters; (2) The objective function of the integrated optimization model for the design and manufacture of the axle bridge is: Total cost C of shaft bridge design and manufacturing 总 , including the cost C S in the design stage and the cost C Z in the manufacturing stage; (3) The constraint functions include: Fatigue life constraint g1(x1, x2, x3, x4, x5, x6), Maximum deformation constraint g2(x1, x2, x3, x4, x5, x6), Maximum stress constraint g3(x1, x2, x3, x4, x5, x6), Maximum deformation constraint g4(x1, x2, x3, x4, x5, x6, x7, x8, x9) when turning the small shaft neck, Maximum stress constraint g5(x1, x2, x3, x4, x5, x6, x7, x8, x9) when turning the small shaft neck. The maximum deformation constraint g6(x1, x2, x3, x4, x5, x6, x 10 , x 11 , x 12 ) and Maximum stress constraint when turning large journal g7(x1,x2,x3,x4,x5,x6,x 10 ,x 11 ,x 12 ); The part inside the brackets of each constraint function is the design parameter, and the part outside the brackets of each constraint function is the function name of the constraint function. Each constraint function is an implicit function of its design parameters. According to Formula 1, that is, the integrated optimization model for the design and manufacture of the axle bridge, the dimensional parameters of the axle bridge and the machining parameters of the axle bridge are optimized by sequential quadratic programming: In Formula 1, to are the thresholds of the corresponding constraint functions determined by the enterprise; Initial values are assigned to each design parameter based on the existing size parameters and machining parameters of the shaft bridge. Simulation iterative calculations are performed according to Formula 1, and different values of each design parameter in the simulation iterative calculations are screened out on the condition that each constraint function satisfies the corresponding threshold, and the C 总 The smallest C 总 The values of each design parameter corresponding to the value are the optimal values of each design parameter; It also includes the total cost C of the shaft bridge design and manufacturing 总 Optimization method: Total cost C of shaft bridge design and manufacturing 总 The calculation formula is C 总 = C S + C Z ; Design phase cost C S = C 材 + C 人 , C 人 is the labor cost in the design phase; C 材 is the material cost; Material cost C 材 Calculated according to Formula 2, and Formula 2 is: Wherein, V is the volume of the forging and welding blank of the shaft bridge, ρ is the material density of the forging and welding blank of the shaft bridge, C 单价 is the unit price of the material of the forging and welding blank of the shaft bridge, l1 is the axial length of the small shaft neck, l2 is the axial length of the large shaft neck; l3 is the length of the shaft body; l4 is the total width of the shaft bridge; l5 is the height of the shaft head; Manufacturing stage cost C of the axle bridge Z Calculated according to Formula 3, and Formula 3 is: C Z = 2 × C 小 + 2 × C 大 + C 其它 ; In Equation 3, C 小 is the total processing cost of the small journal, C 大 is the total processing cost of the large journal, and C 其它 is the processing cost of other shaft bridge sections; The total processing cost C of the small journal 小 includes the direct turning cost C M , the tool change cost C R and the tool wear cost C D , and the corresponding calculation formula is as follows: C M It is determined by the actual cutting amount and the cost coefficient K0 during the machining of the small journal. K0 represents the sum of the labor cost and management cost per unit time. K t represents the tooling cost coefficient for machining the small journal; t e Indicates the tool change time T p represents the tool life l1 represents the length of the small shaft neck. C 其它 including the machining cost of the shaft body and the boring machining cost at the spring seat, The total processing cost C of the large journal 大 including the direct turning cost tool change cost and tool wear cost The calculation formula is as follows: It is determined by the actual cutting amount and the cost coefficient K0 during the machining of the large journal; K0 represents the sum of the labor cost and management cost per unit time. K t * represents the tool cost coefficient for machining large journal t e Indicates the tool change time T p represents the tool life; Construct a double aggregation model of the fatigue life of the axle bridge. In Formula 1, the fatigue life constraint g1(x1, x2, x3, x4, x5, x6) is an implicit function of x1 to x6, and the specific value of g1 is obtained through experiments or finite element simulations. Adopt a dual aggregation model Approximately fit the implicit relationship function between design parameters and fatigue life. The dual aggregation model The specific expression is Formula Four: In Equation 4, is the aggregated Kriging model, is the aggregated RBF model, is the aggregated SVR model; w K is the weight value of the aggregated Kriging model, w R is the weight value of the aggregated RBF model, w S is the weight value of the aggregated SVR model; The construction of Formula Four includes two steps. The first step is to construct and (4.1) The expression of the aggregated Kriging model is specifically: Among them, is a Kriging model with Corrgauss as the correlation model, It is a Kriging model with corrlin as the correlation model. It is a Kriging model with Cubic as the relevant model; The calculation formulas of w1, w2, and w3 in the aggregated Kriging model are: In the calculation formulas of w1, w2 and w3, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; collectively refer to and as correlation coefficient The calculation formula of is as follows: Correlation coefficient In the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft-bridge performance response at the i-th sample point, is the mean value of the true values of the performance responses at n sample points, is the predicted value of the shaft-bridge fatigue life at the i-th sample point, obtained by calling the j-th Kriging approximate model; (4.2) The expression of the aggregated RBF model is specifically: In the expression of the aggregated RBF model, It is an RBF model with Corrgauss as the kernel function. It is an RBF model with linear as the kernel function. It is an RBF model with a Polynomial Kernel as the kernel function; The calculation formulas of w4, w5, and w6 are as follows: In the calculation formulas of w4, w5, and w6, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; The calculation formula of the correlation coefficient is as follows: Coefficient of correlation In the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life values at n sample points, is the predicted value of the shaft bridge fatigue life at the i-th sample point, obtained by calling the j-th RBF approximate model; (4.3) The expression of the aggregated SVR model is specifically as follows: In the expression of the aggregated SVR model, is an SVR model with a linear kernel function, is an SVR model with a PolynomialKernel function, is an SVR model with an RBF kernel function; In the expression of the aggregated SVR model, the calculation formulas for w7, w8, and w9 are as follows: In the calculation formulas of w7, w8, and w9, is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; The calculation formula of the correlation coefficient is as follows: Correlation coefficient In the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft bridge fatigue life at the i-th sample point, is the mean of the true fatigue life values at n sample points, is the predicted value of the shaft bridge fatigue life at the i-th sample point, obtained by calling the j-th SVR approximate model; The second step in constructing Equation Four is to construct a dual aggregation model to obtain Equation Four; In Formula 4, w K , w S , w R are calculated as follows: w K , w S , w R In the calculation formula of is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; Let and be collectively referred to as R 2 . Here, the calculation formula of the correlation coefficient R 2 is as follows: The correlation coefficient R 2 in the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life values at n sample points, is the predicted value of the shaft bridge fatigue life at the i-th sample point, obtained by calling the j-th aggregation model.

2. The optimization method for dimensional parameters and machining parameters according to claim 1, characterized in that: Construct a dual Kriging aggregation model for the maximum deformation and maximum stress in the design and manufacturing stage of the axle bridge The specific expression is Formula Five. The implicit relationship between the axle bridge size parameters, axle bridge machining parameters and performance responses is fitted through Formula Five; Formula Five is: In Equation 5, is the aggregated 0th-order Kriging model, is the aggregated 1st-order Kriging model, is the aggregated 2nd-order Kriging model; w 0K is 's weight value, w 1K is 's weight value, w 2K is 's weight value; The construction of Formula Five includes two steps. The first step is to construct and (5.1) The expression of the aggregated 0th-order Kriging model is: Among them, is a 0th-order Kriging model with Corrgauss as the correlation model, It is a 0th-order Kriging model with corrlin as the correlation model. It is a 0th-order Kriging model with Cubic as the relevant model; The calculation formula for w in the aggregated 0th-order Kriging model 01 , w 02 , w 03 is as follows: w 01 , w 02 , w 03 In the calculation formula of is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; The calculation formula of the correlation coefficient is as follows: Coefficient of correlation In the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle bridge fatigue life at the i-th sample point, obtained through finite element simulation, is the mean value of the true fatigue life values at n sample points, is the predicted value of the axle bridge performance response at the i-th sample point, obtained by calling the j-th 0th-order Kriging approximate model; (5.2) The expression of the aggregated 1st-order Kriging model is specifically: In the expression of the aggregated 1st-order Kriging model, It is a first-order Kriging model with Corrgauss as the correlation model. It is a first-order Kriging model with corrlin as the correlation model. It is a first-order Kriging model with Cubic as the relevant model; w 14 ,w 15 ,w 16 The calculation formula of is as follows: w 14 , w 15 , w 16 In the calculation formula of is 's correlation coefficient, is 's correlation coefficient, is 's correlation coefficient; here, the calculation formula of the correlation coefficient is as follows: Correlation coefficient In the calculation formula of, n is the number of sample points for constructing the approximate model, and y i is the true value of the axle bridge performance response at the i-th sample point, is the predicted value of the axle bridge performance response at the i-th sample point, obtained by calling the j-th first-order Kriging approximate model; (5.3) The expression of the aggregated 2nd-order Kriging model is specifically: In the expression of the aggregated second-order Kriging model, is the second-order Kriging model with Corrgauss as the correlation model, is the second-order Kriging model with corrlin as the correlation model, is the second-order Kriging model with Cubic as the correlation model; w 27 , w 28 , w 29 The calculation formula of is as follows: w 27 , w 28 , w 29 In the calculation formula of is correlation coefficient; The calculation formula of the correlation coefficient is as follows: Correlation coefficient In the calculation formula of, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft-bridge performance response at the i-th sample point, is the mean value of the true values of the performance responses at n sample points, is the predicted value of the shaft-bridge performance response at the i-th sample point, obtained by calling the j-th second-order Kriging approximate model; The second step in constructing Formula Five is to construct a dual aggregation model based on the aggregated models of 0th-order Kriging, 1st-order Kriging, and 2nd-order Kriging Formula Five is obtained; In Formula 5, w 0K , w 1K , w 2K are calculated as follows: w 0K , w 1K , w 2K In the calculation formula of, w 0K is the correlation coefficient of, w 1K is the correlation coefficient of, w 2K is the correlation coefficient of; here the correlation coefficient has the following calculation formula: The correlation coefficient here in the calculation formula, n is the number of sample points for constructing the approximate model, and y i is the true value of the shaft-bridge fatigue life at the i-th sample point, is the mean value of the true fatigue life at n sample points, is the predicted value of the shaft-bridge performance response at the i-th sample point, obtained by calling the j-th aggregation model.

3. The optimization method for dimensional parameters and machining parameters according to claim 2, characterized in that: To construct the mathematical models of each constraint function, sample data is selected to conduct simulation experiments, and the specific steps are as follows: First, 10 candidate sample points are evenly generated throughout the design domain by grid sampling 6 ; In the second step, the given initial design point x0 is used to replace the sample point closest to x0 in the design domain as the first training sample point; In the third step, the distances between the candidate sample points except x0 and the first training sample point x0 are calculated, and the sample point with the largest distance is selected as the second training sample point, designated as x1; Step 4: Calculate the distances between the candidate sample points other than x0 and x1 and x0 and the distance from x1 Select the sample point with the largest distance as the third training sample point, designated as x2; Step 5: Calculate the distances between each candidate sample point and the training sample points x0, x1,..., x k-1 respectively. Select the candidate sample point with the largest value as the (k + 1)-th training sample point; In the sixth step, repeat step five. When the number of training sample points reaches 10n, the sampling ends.

4. The axle bridge manufactured by using the optimization method for dimensional parameters and machining parameters according to any one of claims 1 to 3, characterized in that: The two large journal bearings and the two small journal bearings at the left and right ends of the axle body are coaxially arranged, and this axis is higher than the upper surface of the axle body.

5. The axle bridge according to claim 4, characterized in that: The various parts of the axle bridge are of an integral casting structure. In each unilateral structure, the centers of mass of the large journal bearing and the small journal bearing are located on the same horizontal straight line as the center of mass of the axle head.

6. The axle bridge according to claim 5, characterized in that: The vertical section of the axle head extending in the front-back direction is trapezoidal with a smaller upper part and a larger lower part.

7. The axle bridge according to claim 6, wherein: The bottom surfaces of the spring seat, the axle head, and the axle body are located in the same plane; the left-right width of the spring seat is the same as the left-right width of the axle head.

Images