阅读说明：本技术 一种汽车侧面结构的优化方法 (Optimization method of automobile side surface structure ) 是由 唐樟春 李凌云 夏艳君 刘亚鹏 岳涧洲 周斌 于 2021-06-15 设计创作，主要内容包括：该发明公开了一种汽车侧面结构的优化方法,属于结构优化领域。首先确定优化目标：B柱内侧的厚度x-(1)、B柱加强件的厚度x-(2)、地板内侧的厚度x-(3)、横梁的厚度x-(4)、门梁的厚度x-(5)、车门带线的厚度x-(6)、车顶纵梁的厚度x-(7)、垫B柱内侧的材料x-(8)、垫地板内侧的材料x-(9)；护栏高度x-(10)、护栏撞击的位置x-(11)为非设计变量,为符合N～(0,100)正态分布随机变量,即在-30mm～30mm之间连续变化；再通过确定目标函数和约束条件；最后采用计算最优的优化目标；该方法计算出的优化目标,计算时间更短,优化目标更好。(The invention discloses an optimization method of an automobile side surface structure, and belongs to the field of structure optimization. Firstly, determining an optimization target: thickness x of the inner side of the B-pillar 1 Thickness x of B-pillar reinforcement 2 Thickness x of the inner side of the floor 3 Thickness x of the beam 4 Thickness x of door beam 5 Thickness x of door belt line 6 Thickness x of roof rail 7 Material x inside the pad B column 8 Material x of inner side of floor mat 9 (ii) a Height x of guardrail 10 Position x of guardrail impact 11 Is a non-design variable and is a random variable conforming to N- (0,100) normal distribution, namely continuously changing between-30 mm and 30 mm; determining a target function and a constraint condition; finally miningCalculating an optimal optimization target; the optimization target calculated by the method is shorter in calculation time and better.)

1. a method of optimizing a side structure of an automobile, the method comprising:

step 1: determining the optimization objective includes: thickness x of the inner side of the B-pillar₁Thickness x of B-pillar reinforcement₂Thickness x of the inner side of the floor₃Thickness x of the beam₄Thickness x of door beam₅Thickness x of door belt line₆Thickness x of roof rail₇Material x inside the pad B column₈Material x of inner side of floor mat₉；

Height x of guardrail₁₀Position x of guardrail impact₁₁Is a non-design variable and is a random variable conforming to N- (0,100) normal distribution, namely continuously changing between-30 mm and 30 mm;

step 2: determining an objective function and a constraint condition:

Weight＝1.98+4.9x₁+6.67x₂+6.98x₃+4.01x₄+1.78x₅+2.73x₇

F_Adbom＝1.16-0.3717x₂x₄-0.00931x₂x₁₀-0.484x₃x₉+0.01343x₆x₁₀

Def_{nb_u}＝28.98+3.818x₃-4.2x₁x₂+0.0207x₅x₁₀+6.63x₆x₉-7.7x₇x₈+0.32x₉x₁₀

Def_{nb_m}＝33.86+2.95x₃+0.1792x₁₀-5.057x₁x₂-11x₂x₈-0.0215x₅x₁₀-9.98x₇x₈+22x₈x₉

Def_{nb_l}＝46.36-9.9x₂-12.9x₁x₈+0.1107x₃x₁₀

VC_up＝0.261-0.0159x₁x₂-0.188x₁x₈-0.019x₂x₇+0.0144x₃x₅+0.0008757x₅x₁₀+0.08045x₆x₉+0.00139x₈x₁₁+0.00001575x₁₀x₁₁

VC_mid＝0.214+0.00817x₅-0.131x₁x₈-0.0704x₁x₉+0.03099x₂x₆-0.018x₂x₇+0.0208x₃x₈+0.121x₃x₉-0.00364x₅x₆+0.0007715x₅x₁₀-0.0005354x₆x₁₀+0.00121x₈x₁₁

Vel_B-pillar＝10.58-0.674x₁x₂-1.95x₂x₈+0.02054x₃x₁₀-0.0198x₄x₁₀+0.028x₆x₁₀

in the formula:

weight denotes the total mass of the test model, F_AdbomIndicating the abdominal load, Def, experienced by the test dummy_{nb_u}Representing the amount of deformation of ribs above the chest; def_{nb_m}Representing the deformation of the middle rib of the chest; def_{nb_l}Representing the deformation of the ribs below the chest; VC (vitamin C)_upRepresenting a viscous injury index above the chest; VC (vitamin C)_midRepresenting a viscous lesion index in the middle of the chest; VC (vitamin C)_lowIndicating a viscous lesion index below the chest; force_publicRepresenting pubic symphysis force; vel_B-pillarRepresents the velocity at the midpoint of the B column; vel_doorRepresenting the speed of the front door near the B-pillar;

and step 3: calculating an optimization target with the minimum total mass by adopting a second generation non-dominated sorting evolution algorithm (NSGA-II) to obtain an optimization result;

[1] the crossover operator of the second generation non-dominated sorting evolution algorithm is:

wherein p is_1i，p_2iTwo chromosomal genes, x, being parents in the ith crossover process, respectively_1i，x_2iTwo chromosome genes of filial generation in the ith crossing process respectively;

[2] the mutation operator is:

wherein, Fit_mIn order to be the maximum value of the fitness value,the Fit ═ Weight is the adaptability value of the variant individual as the adaptability mean value of each generation; p_m1，P_m2Respectively an initial mutation probability and a final mutation probability;

when in useWhen is, P_mDecreases with increasing fitness; otherwise, the state is kept unchanged;

[3] the new objective function and fitness function are constructed as follows:

Fit(x_i)＝G(x_i)

in the formula:

wherein: i represents an individual with number i in the population, G (x)_i) New objective function values for the individual; fit (x)_i) Is the corresponding individual fitness value; f (x)_i) Is the corresponding original objective function value; f (x)_i) Is the corresponding penalty function value;is the overall objective function mean;

g_j(x_i) Corresponding to the case of deviation from the constraint condition; m is the number of constraint conditions; α is a penalty factor constant.

Technical Field

The invention belongs to the field of structure optimization, and particularly relates to an optimization design method for an automobile side surface structure.

Background

With the rapid development of the automobile industry, automobiles are more and more popular, and traffic accidents are frequent. Among the car collision accidents, the side pillar collision is a traffic accident that often occurs. The pressure of the side column collision on the vehicle door and the feedback force on the human body are large, and the vehicle door has great harm to the chest of the human body. Therefore, a simulation automobile side column collision test is required during automobile design, and an automobile is improved according to data obtained by the test, so that a safer and more reliable automobile is designed.

However, because the time and the economic cost of the collision test are generally higher, before the collision test is carried out, the computer simulation research has an irreplaceable effect, the times of the collision test can be reduced, the development cost is saved, meanwhile, the virtual collision test can be carried out in the conceptual design stage, and the research and development period is shortened. The invention solves the general mathematical model through the improved parallel genetic algorithm, and provides theoretical basis and reference for the design parameters of key parts of the collision simulation test.

Disclosure of Invention

The purpose of the invention is as follows: in order to overcome the defects in the prior art, the invention provides the automobile side collision structure optimization method for improving the genetic algorithm, so that the efficiency and the precision of the automobile side collision structure optimization are improved.

The technical scheme of the invention is as follows: a method of optimizing a side structure of an automobile, the method comprising:

step 1: determining the optimization objective includes: thickness x of the inner side of the B-pillar₁Thickness x of B-pillar reinforcement₂Thickness x of the inner side of the floor₃Thickness x of the beam₄Thickness x of door beam₅Thickness x of door belt line₆Thickness x of roof rail₇Material x inside the pad B column₈Material x of inner side of floor mat₉；

Height x of guardrail₁₀Position x of guardrail impact₁₁Is a non-design variable and is a random variable conforming to N- (0,100) normal distribution, namely continuously changing between-30 mm and 30 mm;

step 2: determining an objective function and a constraint condition:

Weight＝1.98+4.9x₁+6.67x₂+6.98x₃+4.01x₄+1.78x₅+2.73x₇

F_Adbom＝1.16-0.3717x₂x₄-0.00931x₂x₁₀-0.484x₃x₉+0.01343x₆x₁₀

Def_{nb_u}＝28.98+3.818x₃-4.2x₁x₂+0.0207x₅x₁₀+6.63x₆x₉-7.7x₇x₈+0.32x₉x₁₀

Def_{nb_m}＝33.86+2.95x₃+0.1792x₁₀-5.057x₁x₂-11x₂x₈-0.0215x₅x₁₀ -9.98x₇x₈+22x₈x₉

Def_{nb_l}＝46.36-9.9x₂-12.9x₁x₈+0.1107x₃x₁₀

VC_up＝0.261-0.0159x₁x₂-0.188x₁x₈-0.019x₂x₇+0.0144x₃x₅ +0.0008757x₅x₁₀+0.08045x₆x₉+0.00139x₈x₁₁+0.00001575x₁₀x₁₁

VC_mid＝0.214+0.00817x₅-0.131x₁x₈-0.0704x₁x₉+0.03099x₂x₆-0.018x₂x₇+0.0208x₃x₈ +0.121x₃x₉-0.00364x₅x₆+0.0007715x₅x₁₀-0.0005354x₆x₁₀+0.00121x₈x₁₁

Vel_B-pillar＝10.58-0.674x₁x₂-1.95x₂x₈+0.02054x₃x₁₀-0.0198x₄x₁₀+0.028x₆x₁₀

in the formula:

weight denotes the total mass of the test model, F_AdbomIndicating the abdominal load, Def, experienced by the test dummy_{nb_u}Representing the amount of deformation of ribs above the chest; def_{nb_m}Representing the deformation of the middle rib of the chest; def_{nb_l}Representing the deformation of the ribs below the chest; VC (vitamin C)_upRepresenting a viscous injury index above the chest; VC (vitamin C)_midRepresenting a viscous lesion index in the middle of the chest; VC (vitamin C)_lowIndicating a viscous lesion index below the chest; force_publicRepresenting pubic symphysis force; vel_B-pillarRepresents the velocity at the midpoint of the B column; vel_doorRepresenting the speed of the front door near the B-pillar;

and step 3: calculating an optimization target with the minimum total mass by adopting a second generation non-dominated sorting evolution algorithm (NSGA-II) to obtain an optimization result;

[1] the crossover operator of the second generation non-dominated sorting evolution algorithm is:

wherein p is_1i，p_2iTwo chromosomal genes, x, being parents in the ith crossover process, respectively_1i，x_2iTwo chromosome genes of filial generation in the ith crossing process respectively;

[2] the mutation operator is:

wherein, Fit_mIn order to be the maximum value of the fitness value,the Fit ═ Weight is the adaptability value of the variant individual as the adaptability mean value of each generation; p_m1，P_m2Respectively an initial mutation probability and a final mutation probability;

when in useWhen is, P_mDecreases with increasing fitness; otherwise, the state is kept unchanged;

[3] the new objective function and fitness function are constructed as follows:

Fit(x_i)＝G(x_i)

in the formula:

wherein: i represents an individual with number i in the population, G (x)_i) New objective function values for the individual; fit (x)_i) Is the corresponding individual fitness value; f (x)_i) Is the corresponding original objective function value; f (x)_i) Is the corresponding penalty function value;is the overall objective function mean; g_j(x_i) Corresponding to the case of deviation from the constraint condition; m is the number of constraint conditions; α is a penalty factor constant.

The invention provides a method for carrying out unconstrained processing on a target function by using an internal penalty function method; a cross operator based on normal distribution is adopted; the mutation operator with the adaptive degree change is adopted, and the common test functions, such as an Ackley function, a Sphere function, a Griewank function and the like, are adopted to test the mutation operator, so that the method has high-precision global optimization capability.

Drawings

Fig. 1 is a diagram of an iterative optimization result according to an embodiment of the present invention.

Detailed Description

Step 1: determining the optimization objective includes: thickness x of the inner side of the B-pillar₁Thickness x of B-pillar reinforcement₂Thickness x of the inner side of the floor₃Thickness x of the beam₄Thickness x of door beam₅Thickness x of door belt line₆Thickness x of roof rail₇Material x inside the pad B column₈Material x of inner side of floor mat₉. As in the following table:

design variables for side impact model

Height x of guardrail₁₀Position x of guardrail impact₁₁Is a non-design variable and is a random variable conforming to the normal distribution of N to (0,100), namely continuously changing between-30 mm and 30 mm.

Step 2: determining an objective function and a constraint condition: in a specific implementation case, according to the existing security standards at home and abroad, the actual values of the constraints are as follows:

Weight＝1.98+4.9x₁+6.67x₂+6.98x₃+4.01x₄+1.78x₅+2.73x₇

F_Adbom＜1.0KN

Def_{nb_u}＜32mm

Def_{nb_m}＜32mm

Def_{nb_l}＜32mm

VC_up＜0.32m/s

VC_mid＜0.32m/s

VC_low＜0.32m/s

Force_public＜4.0KN

Vel_B-pillar＜9.9mm/ms

Vel_door＜15.7mm/ms

in the formula:

weight — total mass of the test model;

F_Adbom-abdominal loading of the test dummy;

Def_{nb_u}-the amount of deformation of the ribs above the chest;

Def_{nb_m}-the amount of deformation of the middle ribs of the chest;

Def_{nb_l}-the amount of deformation of the ribs below the chest;

VC_up-an indication of adhesive injury over the chest;

VC_mid-middle of chest viscous injury index;

VC_low-an indication of adhesive injury below the chest;

Force_public-pubic symphysis force;

Vel_B-pillar-velocity at the midpoint of the B column;

Vel_door-speed of the front door near the B-pillar;

and step 3: on the basis of a non-dominated sorting genetic algorithm (NSGA-II), certain improvement is carried out on a crossover operator and a mutation operator of genetic operation by combining the characteristics of structure optimization design.

1. Improvements in crossover operators

The NSGA-II algorithm typically uses a binary crossover operator to perform a single-point crossover of a chromosome gene, as shown in the following formula:

wherein the content of the first and second substances,

p_1i，p_2itwo chromosome genes of a parent in the ith crossing process respectively;

x_1i，x_2itwo chromosome genes of filial generation in the ith crossing process respectively;

α is represented by the formula:

in the formula:

β is a random number uniformly distributed over (0, 1);

γ is a non-negative number defined by the decision maker.

The gamma value selection has the characteristics of subjectivity and randomness, and the stability of the algorithm cannot be guaranteed, so that the defect can be effectively overcome by adopting a normal distribution-based crossover operator (NDX). As shown in the following formula:

in the formula:

l N (0,1) | -a normally distributed random variable;

u-random number on (0, 1).

2. Improvement of mutation operator

The mutation operator is proposed to avoid the optimization process being limited by a certain region, and the individual can jump out of the original region for optimization through mutation operation, so that not only is the calculation efficiency improved, but also the diversity of understanding is kept. In the mutation process, a self-adaptive mutation concept is introduced on the basis of polynomial mutation, and an operator adjusts the mutation probability along with the fitness so that the leading edge distribution of the optimal solution set is more reasonable as shown in the following formula:

wherein the content of the first and second substances,

Fit_mis the maximum fitness value;

is the fitness mean value of each generation; wherein F (x), g_i(x) Is a continuous function

The feasible field is Fit ═ -Weight is the fitness value of the variant individual;

P_m1，P_m2respectively an initial mutation probability and a final mutation probability;

when in useWhen is, P_mDecreases with increasing fitness; otherwise, the value is kept unchanged.

3. Improvements in constraints

A plurality of constraint conditions are attached to the traditional structure optimization problem, so that the genetic optimization process is more complex, the solving difficulty is increased, and the problem is processed by adopting an internal penalty function method. The overall idea of the method is as follows: setting a new function, the definition domain of the function is the value range of the constraint condition, then multiplying the function by a penalty coefficient, when the variable approaches the constraint boundary, the function value tends to infinity, and then adding the original function to the function to form a new function.

For the structural dimension optimization problem:

min F(x)

s.t.g_i(x)≤0,i＝1,...,m

wherein F (x), g_i(x) Is a continuous function

Can be operated in

S＝{x|g_i(x)≤0,i＝1,...,m}

According to the idea of a penalty function method, establishing an unconstrained objective function in a new definition domain as shown in the specification:

G(x,r)＝F(x)+rf(x)s.t.x∈intS

wherein f (x) is a continuous function when x → g_i(x) When bounding, f (x) → + ∞.

f (x) the usual expression form is:and

r is a small positive number. When x → g_i(x) When constraining the boundary, f (x) → + ∞, the corresponding function value for x must not be the minimum value. Otherwise, rf (x) → 0, G (x, r) ≈ f (x).

According to the idea of a Lemonge penalty function, a method for punishing an infeasible solution according to the deviation condition of a constraint condition is adopted, so that the condition that an individual close to the inner side of the constraint condition boundary is punished can be avoided, the infeasible solution is also punished reasonably, the optimization capability of the algorithm is improved, and a new objective function and a fitness function are constructed as follows:

in the formula:

wherein: i represents an individual with number i in the population, G (x)_i) New objective function values for the individual; fit (x)_i) Is the corresponding individual fitness value; f (x)_i) Is the corresponding original objective function value; f (x)_i) Is the corresponding penalty function value;is the overall objective function mean; g_j(x_i) Corresponding to the case of deviation from the constraint condition; m is the number of constraint conditions; α is a penalty factor constant, here taken to be 2. The results of SQP, SA versus improved NSGA-II optimization are shown in the following table: