Automobile side collision optimization design method based on coevolution algorithm

Abstract

The invention discloses a coevolution algorithm-based automobile side collision optimization design method, which comprises the following steps of: firstly, constructing a passenger injury model, an automobile body structure deformation model and an automobile body weight model based on an injury mechanism in a collision accident, and providing a basis for subsequent optimization; through a multi-objective optimization model, objectives of passenger safety, a vehicle body structure, lightweight design and the like are considered, and strict constraint conditions are set. And a coevolution algorithm is utilized, and coevolution of the main population and the auxiliary population is combined. By combining the constraint processing technology, it is ensured that the design scheme meets the actual collision safety requirement. And finally, through optimization, an optimal automobile structure design scheme which not only meets the safety requirement but also has good performance can be output.

Description

Technical Field

[0001] The present invention belongs to the technical field of automotive safety, and particularly relates to an optimization design method for automotive side collision based on co-evolution algorithm. Background Art

[0002] Attention has been paid. Especially in automotive side collision accidents, the design of the body structure, energy absorption capacity, and occupant protection performance are directly related to the degree of injury and survival rate during the accident. Therefore, improving the safety of automotive side collision and optimizing the body structure have become the core issues that need to be solved urgently in modern automotive design.

[0003] Currently, the optimization design of automotive side collision mainly relies on traditional finite element analysis and experimental data analysis. By simulating the side collision process, analyzing the force condition and deformation mode of the body during the collision, an ideal structure can be designed. Although these methods can provide certain theoretical basis, they have the following defects: traditional collision optimization design often only focuses on one aspect of the goal, such as energy absorption efficiency or structural strength, ignoring the balance of multiple goals. In practical applications, designers need to find the best compromise among multiple goals, such as vehicle body weight, structural stiffness, and safety, etc. In the actual automotive collision optimization process, multiple complex constraints that are interrelated are usually encountered, such as body shape, material properties, and manufacturing processes, etc. These constraint conditions are often difficult to accurately quantify at the initial stage of design, and traditional methods are difficult to handle multiple constraints simultaneously, resulting in the optimization design results not being practically feasible.

[0004] Therefore, how to take into account factors such as lightweight of the body structure, cost control, and manufacturability while ensuring the safety performance of the vehicle has become a major challenge faced by the automotive industry. Summary of the Invention

[0005] The purpose of the present invention is to provide an optimization design method for automotive side collision based on co-evolution algorithm, which solves the problem of single-objective optimization existing in the prior art.

[0006] The technical solution adopted by the present invention is an optimization design method for automotive side collision based on co-evolution algorithm, which is specifically implemented according to the following steps:

[0007] Step 1, construct an occupant injury model;

[0008] Step 2, establish an automotive structure deformation model;

[0009] Step 3, construct a vehicle body weight model;

[0010] Step 4, construct a multi-objective optimization problem model and set corresponding constraint conditions;

[0011] Step 5: Execute the co-evolution algorithm to solve the constrained multi-objective optimization problem and optimize it to the unconstrained Pareto front.

[0012] Step 6: Optimize by combining constraint handling techniques. Through the constraint handling techniques and in combination with specific conditions of vehicle collisions, optimize it to the constrained Pareto front; determine whether Step 7 can be executed, otherwise return to Step 5.

[0013] Step 7: Finally, output the vehicle structure design solution that meets the constraint conditions.

[0014] The features of the present invention also lie in:

[0015] In Step 1, construct an occupant injury model, comprehensively consider head injury and the peak chest acceleration, and the goal is to minimize these two types of injuries, specifically as follows:

[0016] min f1(x) = w1·HIC(x) + w2·Acc peak (x) (1)

[0017] Among them, w1 and w2 represent weight coefficients, reflecting the importance of HIC and Acc peak ;

[0018] The calculation formula of HIC(x) is:

[0019]

[0020] Among them, t2 - t1 ≤ 0.036s, a head represents the head acceleration;

[0021] Acc peak (x) The calculation formula is:

[0022]

[0023] Among them, a chest represents the chest acceleration.

[0024] In Step 2, establish a vehicle structure deformation model. The deformation of the vehicle body is determined by elastic deformation and plastic deformation, and the goal is to reduce the maximum deformation:

[0025] min f2(x) = max(δ elastic , δ plastic ) (4)

[0026] Elastic deformation displacement:

[0027]

[0028] Plastic deformation displacement:

[0029]

[0030] Among them, F elastic and E plastic represent the acting forces in the elastic and plastic stages respectively, L is the length of the deformation area, A is the cross-sectional area of the deformation area, and E plastic and E plastic represent the elastic modulus and plastic modulus of the material respectively.

[0031] In step 3, a vehicle body weight model is constructed, and the vehicle body weight is reduced by optimizing materials and designs:

[0032]

[0033] Among them, represents the density of the i-th material, and V i represents the volume of the i-th material.

[0034] In step 4, a constrained multi-objective optimization problem model for optimizing the side collision of the vehicle is constructed. Constraint conditions such as material cost and manufacturing process are set. At the same time, the optimization objectives include reducing the peak occupant acceleration and optimizing energy absorption. The constraint conditions are as follows:

[0035] Constraint condition 1: Occupant safety limit

[0036] The head injury criterion must be less than the maximum value specified by the safety standard:

[0037] g1(x) = HIC(x) - HIC max ≤0 (8)

[0038] The peak chest acceleration shall not exceed the maximum allowable value:

[0039] g2(x) = Acc peak (x) - Acc max ≤0 (9)

[0040] Constraint condition 2: Vehicle body deformation limit

[0041] The elastic deformation shall not exceed the limit value:

[0042] g3(x) = δ elastic -δ elastic，max ≤0 (10)

[0043] The plastic deformation shall not exceed the limit value:

[0044] g4(x) = δ plastic -δ plastic，max ≤0 (11)

[0045] Constraint condition 3: Vehicle size limit

[0046] Width limit:

[0047] g5(x) = x width -x width，max ≤0 (12)

[0048] Height limit:

[0049] g6(x) = x height -x height，max ≤0 (13)

[0050] Length limit:

[0051] g7(x) = x length -x length，max ≤0 (14)

[0052] Constraint 4: Cost limit

[0053]

[0054] where c i represents the cost of the i-th design parameter, x i represents the i-th design variable, and C max represents the cost upper limit;

[0055] Constraint 5: Vehicle stability after collision

[0056] Roll angle limit after collision:

[0057] g9(x) = Roll Angle(x) - Roll Angle max ≤0 (16)

[0058] Yaw angle limit after collision:

[0059] g 10 (x) = Yaw Angle(x) - Yaw Angle max ≤0 (17)

[0060] In steps 1 - 4, the optimization problem can be summarized as:

[0061] Minimize f(x) = (f1(x), f2(x), f3(x)) (18)

[0062] subject to: g i (x) ≤ 0, i = 1, 2,..., 10 (19)

[0063] where f1(x) represents the degree of occupant injury, f2(x) represents the vehicle body deformation, f3(x) represents the vehicle weight, and g i (x) represents the constraint condition.

[0064] Step 5 is specifically implemented according to the following steps:

[0065] Step 5.1: Initialize the individual of the vehicle side collision parameters to generate two initial populations P1 and P2, where each individual in the population corresponds to a possible solution;

[0066] Step 5.2, objective function evaluation. For the parameter individuals in the population, calculate their objective values under all objective functions, and perform constraint processing to evaluate whether the individuals meet the constraint conditions in the problem. If an individual meets all the constraint conditions, add it to the archive. When the archive is full, sort all the feasible solutions found, and select the top N optimal individuals to add to the archive;

[0067] Step 5.3, Population P1 generates offspring using differential evolution, and population P2 generates offspring using genetic algorithms. The two populations update the archive A respectively according to their own populations and the combined offspring. Populations P1 and P2 update the next generation respectively according to their own populations and the offspring;

[0068] Step 5.4, when the population converges to the unconstrained Pareto front, output the final population P and the archive A. Population P is output by combining P1 and P2 and selecting the optimal N solutions.

[0069] In Step 5.1, in order to store and save the feasible solutions found during the evolution process, a file set is constructed, which records the better feasible solutions in the current population and the historical populations.

[0070] For the initialization of the population size N, the formula is as follows:

[0071] P = {x1, x2,..., x N} (20)

[0072] where the decision vector x i = (x i1 , x i2 ,..., x id ) of each individual is generated from the solution space by uniform random sampling:

[0073]

[0074] where x ij represents the jth decision variable of the ith individual, and is uniformly selected from the interval [L j , U j . L j and U j are the lower and upper bounds of the decision variable x i ;

[0075] The archive is used to record the feasible solutions found during the convergence process towards the unconstrained Pareto front, so as to ensure that the algorithm can store and track the optimal solution set in multi-objective optimization. At initialization, the archive is empty and its size is N:

[0076] In step 5.1, to store and preserve the feasible solutions found during the evolution process, an archive set is constructed, which records the better feasible solutions in the current population and the historical populations.

[0077] For the initialization of the population size N, the formula is as follows:

[0078] P = {x1, x2, …, x N} (20)

[0079] where the decision vector x i = (x i1 , x i2 , …, x id ) of each individual is generated from the solution space by uniform random sampling:

[0080]

[0081] where x ij represents the j-th decision variable of the i-th individual, and is uniformly selected from the interval [L j , U j , and L j and U j are the lower and upper bounds of the decision variable x i ;

[0082] The archive is used to record the feasible solutions found during the convergence process towards the unconstrained Pareto front, so as to ensure that the algorithm can store and track the optimal solution set in multi-objective optimization. At initialization, the archive is empty and its size is N:

[0083]

[0084] Step 6 is specifically implemented according to the following steps:

[0085] Step 6.1, initialize the parameter individuals, use the population P output in step 5 as the initial population of the auxiliary population in step 6, and use the output archive A as the initial population of the main population;

[0086] Step 6.2, merge the main population and the auxiliary population, and generate offspring using the genetic algorithm;

[0087] Step 6.3, apply the biased penalty strategy in the main population to punish the infeasible solutions found, and apply the standard deviation-guided auxiliary population perturbation strategy in the auxiliary population to adjust the search direction of the auxiliary population; the calculation formula of the biased penalty strategy is as follows:

[0088]

[0089] Among them, F(x i ) and F(x i ) respectively represent the target vectors of individual x i before and after correction. is a normalized form of the constraint violation degree, where CV min represents the solution with the minimum constraint violation degree in the current solution set, and CV max represents the solution with the most constraint violations in the current solution set. is composed of the minimum value of each objective in the target vectors within the sub-region associated with F(x i ).

[0090] The standard deviation-guided auxiliary population perturbation strategy dynamically adjusts the perturbation intensity by calculating the standard deviation difference between the main population and the auxiliary population. The specific perturbation calculation formula is as follows:

[0091]

[0092] Through formula (24), a perturbation term based on the standard deviation can be added to achieve the purpose of exploring the solution space near the infeasible solutions.

[0093] In the perturbation term, the calculation formula for the perturbation intensity σ' is:

[0094]

[0095] Among them, N(0, 1) represents a normal distribution with a mean of 0 and a variance of 1. The introduction of this distribution ensures the randomness and continuity of the perturbation term, which helps to improve the diversity of solutions. θ1 represents the standard deviation of the main population, and θ2 represents the standard deviation of the auxiliary population. The specific calculation formula for the θ standard deviation is as follows:

[0096]

[0097] Among them, N is the number of populations, f j，i represents the function value of the i-th individual on the j-th objective, represents the average value of the j-th objective function, and M represents the objective dimension.

[0098] Step 6.4, the main population and the auxiliary population combine to generate offspring, and N solutions are selected according to the fitness ranking to enter the next generation.

[0099] The beneficial effects of the present invention are

[0100] The vehicle side collision optimization design method based on the co-evolution algorithm of the present invention can provide significant advantages in multi-objective optimization problems by means of the co-evolution of double populations and dynamic archive update using the co-evolution algorithm. It can not only balance global and local searches, ensure the diversity of solutions, but also effectively handle complex constraint conditions, and finally generate an optimized design solution that meets the actual requirements. In the vehicle side collision problem, the co-evolution algorithm can provide more reliable, feasible and diverse solutions for vehicle design through multi-objective optimization and constraint handling techniques, meeting multiple requirements such as safety, performance and weight. Brief Description of the Drawings

[0101] Figure 1 It is a schematic flow diagram of the vehicle side collision optimization design method based on the co-evolution algorithm of the present invention. Detailed Embodiment

[0102] The present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0103] The present invention provides a vehicle side collision optimization design method based on the co-evolution algorithm, as Figure 1 shown, and is specifically implemented according to the following steps:

[0104] Step 1, construct an occupant injury model;

[0105] Step 2, establish a vehicle structure deformation model;

[0106] Step 3, construct a vehicle body weight model;

[0107] Step 4, construct a multi-objective optimization problem model and set corresponding constraint conditions;

[0108] Step 5, execute the co-evolution algorithm, solve the constrained multi-objective optimization problem, and optimize to the unconstrained Pareto front;

[0109] Step 6, perform optimization in combination with the constraint handling technique, through the constraint handling technique, and in combination with the specific conditions of vehicle collision, optimize to the constrained Pareto front; judge whether step 7 can be executed, otherwise return to step 5;

[0110] Step 7, finally output a vehicle structure design solution that meets the constraint conditions.

[0111] Embodiment 1

[0112] For the vehicle side collision optimization design method based on the co-evolution algorithm, in step 1, when constructing the occupant injury model, the head injury (HIC) and the peak chest acceleration (Acc peak ) are comprehensively considered, and the goal is to minimize the two types of injuries, specifically as follows:

[0113] min f1(x) = w1·HIC(x) + w2·Acc peak (x) (1)

[0114] Wherein, w1 and w2 represent weight coefficients, reflecting the importance of HIC and Acc peak ;

[0115] The calculation formula of HIC(x) is:

[0116]

[0117] Wherein, t2 - t1 ≤ 0.036s, a head represents the head acceleration;

[0118] Acc peak (x) The calculation formula is:

[0119]

[0120] Wherein, a chest represents the chest acceleration.

[0121] Example 2

[0122] An optimized design method for vehicle side collision based on co - evolutionary algorithm. In step 2, a vehicle structure deformation model is established. The vehicle body deformation is determined by elastic deformation and plastic deformation, and the goal is to reduce the maximum deformation:

[0123] minf2(x) = max(δ elastic , δ plastic ) (4)

[0124] Elastic deformation displacement:

[0125]

[0126] Plastic deformation displacement:

[0127]

[0128] Wherein, F elastic , E plastic respectively represent the acting forces in the elastic and plastic stages, L is the length of the deformation region, A is the cross - sectional area of the deformation region, E plastic , E plastic respectively represent the elastic modulus and plastic modulus of the material.

[0129] Example 3

[0130] An optimized design method for vehicle side collision based on co - evolutionary algorithm. In step 3, a vehicle body weight model is constructed, and the vehicle body weight is reduced by optimizing materials and design:

[0131]

[0132] Among them, represents the density of the i-th material, and V i represents the volume of the i-th material.

[0133] Example 4

[0134] An optimized design method for vehicle side collisions based on a co-evolutionary algorithm. In step 4, a constrained multi-objective optimization problem model for vehicle side collision optimization is constructed, and constraint conditions such as material cost and manufacturing process are set. At the same time, the optimization objectives include reducing the peak occupant acceleration and optimizing energy absorption. The constraint conditions are as follows:

[0135] Constraint condition 1: Occupant safety limit

[0136] The head injury criterion must be less than the maximum value specified by the safety standard:

[0137] g1(x) = HIC(x) - HIC max ≤ 0 (8)

[0138] The peak chest acceleration must not exceed the maximum allowable value:

[0139] g2(x) = Acc peak (x) - Acc max ≤ 0 (9)

[0140] Constraint condition 2: Vehicle body deformation limit

[0141] The elastic deformation must not exceed the limit value:

[0142] g3(x) = δ elastic -δ elastic,max ≤ 0 (10)

[0143] The plastic deformation must not exceed the limit value:

[0144] g4(x) = δ plastic -δ plastic,max ≤ 0 (11)

[0145] Constraint condition 3: Vehicle size limit

[0146] Width limit:

[0147] g5(x) = x width -x width,max ≤ 0 (12)

[0148] Height limit:

[0149] g6(x) = x height -xheight,max ≤0 (13)

[0150] Length limit:

[0151] g7(x) = x length -x length,max ≤0 (14)

[0152] Constraint 4: Cost limit

[0153]

[0154] where c i represents the cost of the i-th design parameter, x i represents the i-th design variable, and C max represents the cost upper limit;

[0155] Constraint 5: Vehicle stability after collision

[0156] Roll angle limit after collision:

[0157] g9(x) = Roll Angle(x) - Roll Angle max ≤0 (16)

[0158] Yaw angle limit after collision:

[0159] g 10 (x) = Yaw Angle(x) - Yaw Angle max ≤0 (17)

[0160] In steps 1 - 4, the optimization problem can be summarized as:

[0161] Minimize f(x) = (f1(x), f2(x), f3(x)) (18)

[0162] subject to: g i (x) ≤ 0, i = 1, 2, …, 10 (19)

[0163] where f1(x) represents the degree of occupant injury, f2(x) represents the vehicle body deformation, f3(x) represents the vehicle weight, and g i (x) represents the constraint condition.

[0164] Example 5

[0165] An optimization design method for vehicle side collision based on the co - evolutionary algorithm, where in step 5, step 5 is specifically implemented according to the following steps:

[0166] Step 5.1: Initialize the individual parameters of the vehicle side collision, generate two initial populations P1 and P2, where each individual in the population corresponds to a possible solution; to store and preserve the feasible solutions found during the evolution process, construct an archive set that records the better feasible solutions in the current and historical populations.

[0167] For the initialization of the population size N, the formula is as follows:

[0168] P = {x1, x2,..., x N} (20)

[0169] where the decision vector x i = (x i1 , x i2 ,..., x id ) of each individual is generated from the solution space through uniform random sampling:

[0170]

[0171] where x ij represents the jth decision variable of the ith individual, and is uniformly selected from the interval [L j , U j , where L j and U j are the lower and upper bounds of the decision variable x i ;

[0172] The archive is used to record the feasible solutions searched during the convergence process towards the unconstrained Pareto front, to ensure that the algorithm can store and track the optimal solution set in multi-objective optimization. At initialization, the archive is empty and has a size of N:

[0173]

[0174] Step 5.2, Objective function evaluation. For the parameter individuals in the population, calculate their objective values under all objective functions, perform constraint handling to evaluate whether the individuals meet the constraint conditions in the problem. If an individual meets all the constraint conditions, add it to the archive. When the archive is full, sort all the searched feasible solutions and select the top N optimal individuals to add to the archive;

[0175] Step 5.3, The population P1 generates offspring using differential evolution, and the population P2 generates offspring using genetic algorithms. The two populations update the archive A respectively according to their own populations and the combined offspring. The populations P1 and P2 update the next generation respectively according to their own populations and the offspring;

[0176] Step 5.4, If the current iteration count reaches the iteration stop requirement, transfer to step 5.5, otherwise transfer to step 5.2;

[0177] Step 5.5, when the population converges to the unconstrained Pareto front, output the final population P and the archive A. The population P is obtained by merging P1 and P2 and selecting the optimal N solutions for output.

[0178] Example 6

[0179] For the vehicle side impact optimization design method based on the co-evolution algorithm, in step 5, the method of combining two methods of generating offspring in the dual population with the archive update technique can effectively improve the optimization effect of the vehicle side impact problem in multiple aspects. This method enhances the global and local search capabilities, improves the diversity of solutions, accurately approximates the Pareto front, and effectively handles the constraint conditions, providing a more efficient and reliable optimization process. Finally, it can find a balance among multiple conflicting objectives and complex constraint conditions, so as to output the optimal vehicle structure design scheme with diversity and meeting the actual collision safety requirements. The specific pseudocode is as follows:

[0180]

[0181]

[0182] The pseudocode of the archive-population co-evolution competition strategy is shown in Algorithm 1. The algorithm first generates the initial population and the archive A, both with the size set to N (lines 1 - 2 of Algorithm 1). In each generation of iteration, the algorithm first selects parents from the population to generate offspring through DE (line 5 of Algorithm 1), then selects parents from the population to generate offspring using GA (line 6 of Algorithm 1), and updates the archive A (line 7 of Algorithm 1). The population combines its own offspring to generate the next generation respectively (lines 8 - 9 of Algorithm 1). When the number of evaluations is equal to 50% of the total number of evaluations, based on the fitness, select the optimal N solutions from the population and as the new population P t (lines 10 - 12 of Algorithm 1). The algorithm continuously iterates and updates the population and the archive until the termination condition is met (lines 3 - 14 of Algorithm 1). Finally, the C-TSDP algorithm returns the optimized population P t and the archive A, that is, the final solution set and the archive result (line 15 of Algorithm 1). Through the two mechanisms of generating offspring and the strategy of co-updating the archive and competitively updating the output population in the dual population, the algorithm achieves a balance between the diversity of the population and global convergence during the optimization process.

[0183] Example 7

[0184] Automobile side collision optimization design method based on co-evolution algorithm, wherein step 6 is specifically implemented according to the following steps:

[0185] Step 6.1, initialize the parameter individuals. Use the population P output in step 5 as the initial population of the auxiliary population in step 6, and use the output archive A as the initial population of the main population;

[0186] Step 6.2, merge the main population and the auxiliary population, and generate offspring using the genetic algorithm;

[0187] Step 6.3, apply a biased penalty strategy in the main population to punish the infeasible solutions found, and apply a standard deviation-guided auxiliary population perturbation strategy in the auxiliary population to adjust the search direction of the auxiliary population; The biased penalty strategy dynamically adjusts by punishing all infeasible solutions and, on this basis, adopting a penalty method that favors solutions with better objective values in the sub-region. By dynamically adjusting the penalty strength for infeasible solutions, it ensures that the algorithm can flexibly balance objective optimization and constraint satisfaction at different search stages. This adaptive mechanism can gradually guide the population to converge to the feasible solution space while maintaining the optimization driving force for the objective function. The calculation formula of this strategy is as follows:

[0188]

[0189] Where, F(x i ) and F(x i )’ respectively represent the objective vectors before and after the correction of the individual x i , is a normalized form of the constraint violation degree, where CV min represents the solution with the minimum constraint violation degree in the current solution set, CV max represents the solution with the most constraint violations in the current solution set, is composed of the minimum values of each objective in the objective vectors within the sub-region associated with F(x i );

[0190] The standard deviation-guided auxiliary population perturbation strategy dynamically adjusts the perturbation intensity by calculating the standard deviation difference between the main population and the auxiliary population, thereby guiding the search direction of the auxiliary population to maintain an appropriate difference from the main population. The distribution difference between the two populations is measured by the differentiation of the standard deviation, thus avoiding excessive overlap during the search process. Specifically, when the standard deviation difference between the two populations is large, it indicates that there are significant differences in their distributions in the search space. At this time, by reducing the perturbation intensity, the main population and the auxiliary population can maintain their current states and continue to search in their respective explored regions, thereby avoiding the overlap of search directions. Conversely, when the standard deviation difference between the two populations is small, it means that their search directions tend to be consistent. At this time, by increasing the perturbation intensity, the diversity of the auxiliary population can be enhanced, promoting it to explore new regions to improve the overall search efficiency. This strategy can effectively maintain population diversity, prevent the main population and the auxiliary population from being overly similar during the search process, thereby enhancing the global search ability of the algorithm, improving its ability to jump out of local optimal solutions, and ultimately helping the algorithm find better solutions in complex problems. The specific perturbation calculation formula is as follows:

[0191]

[0192] Through formula (24), it is possible to add a perturbation term based on the standard deviation to achieve the purpose of exploring the solution space near the infeasible solution;

[0193] In the perturbation term, the calculation formula for the perturbation intensity σ is:

[0194]

[0195] where N(0,1) represents a normal distribution with a mean of 0 and a variance of 1. The introduction of this distribution ensures the randomness and continuity of the perturbation term, which helps to improve the diversity of solutions. θ1 represents the standard deviation of the main population, and θ2 represents the standard deviation of the auxiliary population. The specific calculation formula for the θ standard deviation is as follows:

[0196]

[0197] where N is the number of populations, f j,i represents the function value of the i-th individual on the j-th objective, represents the average value of the j-th objective function, and M represents the objective dimension;

[0198] Step 6.4, the main population and the auxiliary population combine with the offspring, and select N solutions according to the fitness ranking to enter the next generation.

[0199] In Step 6, through the co - evolution of combining the main population and the auxiliary population, and by applying the biased penalty strategy and the standard - deviation - guided perturbation strategy, the diversity of solutions, constraint - handling ability, convergence speed, and reliability of the final solution in the optimization process are effectively improved. In the multi - objective optimization of the vehicle side - impact problem, this strategy can not only handle complex constraint conditions but also find a balance among multiple objectives, ensuring that the final output design scheme meets the safety requirements and has good performance and reliability. The specific pseudocode is as follows:

[0200]

[0201]

[0202] In the initialization process of the second stage of the algorithm, first, the solutions in the output population of the first stage are used as the initial solutions of the auxiliary population in the second stage, and the solutions in the output archive are used as the initial solutions of the main population (lines 1 - 5 of Algorithm 2). Then, the main population and the auxiliary population are merged to form a combined population CP t (line 6 of Algorithm 2). Subsequently, according to the sorting of solutions, binary tournament selection is applied to select the solutions with higher fitness from the combined population as parents to construct a mating pool of size 2N (line 7 of Algorithm 2). Then, parent individuals are selected from the mating pool and offspring are generated to form a new solution set O t (line 8 of Algorithm 2). After generating the offspring, first, the main population and the offspring are merged into a set and the dynamic penalty strategy for infeasible solutions is used to dynamically penalize the infeasible solutions in, and then N solutions with higher fitness are selected to enter the next generation (lines 9 - 10 of Algorithm 2). Next, the auxiliary population and the offspring are merged into a set the standard - deviation - guided infeasible - solution perturbation strategy is used to mutate the infeasible solutions in it, and the first N solutions are selected according to the fitness sorting to enter the next generation (lines 11 - 12 of Algorithm 2). Finally, the populations are merged and the first N solutions with high fitness are selected as the output population P t (line 13 of Algorithm 2). Iterate the above steps until the termination condition is met (lines 6 - 13 of Algorithm 2), and finally, the optimized population P is output t (line 15 of Algorithm 2). Through this strategy, the algorithm can effectively utilize the search results of the first stage in the second stage and further approach the Pareto optimal front.

[0203] Based on the above, the present invention effectively solves the problem of vehicle side collision through multi-objective optimization, co-evolutionary algorithm, and constraint handling technology. First, an occupant injury model, a vehicle body structure deformation model, and a vehicle body weight model based on the injury mechanism in collision accidents are constructed, providing a basis for subsequent optimization. Through the multi-objective optimization model, objectives such as occupant safety, vehicle body structure, and lightweight design are considered, and strict constraint conditions are set. Using the co-evolutionary algorithm, by combining the co-evolution of the main population and the auxiliary population, it can not only effectively handle objective conflicts, but also maintain the diversity of solutions and improve the search efficiency.

[0204] During the optimization process of the present invention, by combining constraint handling technology, it is ensured that the design scheme meets the actual collision safety requirements. Finally, after optimization, an optimal vehicle structure design scheme that meets both safety requirements and has good performance can be output. The present invention has comprehensiveness, innovation, and strong practical application value.

Claims

1. An optimization design method for vehicle side collision based on co-evolution algorithm, characterized in that, The implementation is specifically carried out in the following steps: Step 1: Construct an occupant injury model; Step 2: Establish a vehicle structure deformation model; Step 3: Construct a vehicle body weight model; Step 4: Construct a multi-objective optimization problem model and set corresponding constraint conditions; Step 5: Execute a co-evolutionary algorithm to solve the constrained multi-objective optimization problem and optimize it to the unconstrained Pareto front; Step 6: Optimize by combining constraint handling techniques. Through the constraint handling techniques, combined with specific conditions of vehicle collisions, optimize it to the constrained Pareto front; Determine whether Step 7 can be executed, otherwise return to Step 5; Step 7: Finally, output a vehicle structure design solution that meets the constraint conditions.

2. The method for optimizing the design of a vehicle side collision based on a co-evolutionary algorithm according to claim 1, wherein In Step 1, when constructing the occupant injury model, comprehensively consider head injury and the peak chest acceleration. The goal is to minimize these two types of injuries, specifically as follows: min f1(x) = w1·HIC(x) + w2·Acc peak (x) (1) Among them, w1 and w2 represent weight coefficients, reflecting the importance of HIC and Acc peak ; The calculation formula for HIC(x) is: where t2 - t1 ≤ 0.036 s, a head represents the head acceleration; Acc peak (x) The calculation formula is: Among them, a chest represents chest acceleration.

3. The method for optimizing the design of a vehicle side collision based on a co-evolution algorithm according to claim 1, wherein In Step 2, when establishing the vehicle structure deformation model, the vehicle body deformation is determined by elastic deformation and plastic deformation. The goal is to reduce the maximum deformation: min f2(x) = max(δ elastic , δ plastic ) (4) Elastic deformation displacement: Plastic deformation displacement: Among them, F elastic , F plastic respectively represent the acting forces in the elastic and plastic stages, L is the length of the deformation region, A is the cross-sectional area of the deformation region, E eladtic , E plastic respectively represent the elastic modulus and plastic modulus of the material.

4. The vehicle side collision optimization design method based on the co-evolutionary algorithm according to claim 1, wherein In Step 3, when constructing the vehicle body weight model, reduce the vehicle body weight by optimizing materials and design: Among them, represents the density of the i-th material, and V i represents the volume of the i-th material.

5. The optimized design method for vehicle side collision based on co-evolution algorithm according to claim 1, characterized in that In Step 4, construct a constrained multi-objective optimization problem model for optimizing vehicle side collisions, set constraint conditions such as material cost and manufacturing process, and at the same time the optimization objectives include reducing the peak occupant acceleration and optimizing energy absorption. The constraint conditions are as follows: Constraint condition 1: Occupant safety limit The head injury criterion must be less than the maximum value specified by the safety standard: g1(x) = HIC(x) - HIC max ≤ 0 (8) The peak chest acceleration must not exceed the maximum allowable value: g2(x) = Acc peak (x) - Acc max ≤ 0 (9) Constraint condition 2: Vehicle body deformation limit The elastic deformation must not exceed the limit value: g3(x) = δ elastic -δ elastic，max ≤0 (10) The plastic deformation must not exceed the limit value: g4(x) = δ plastic -δ plastic，max ≤ 0 (11) Constraint condition 3: Vehicle size limit Width limit: g5(x) = x width -x width，max ≤ 0 (12) Height limit: g6(x) = x height -x height，max ≤0 (13) Length limit: Constraint condition 4: Cost limit Among them, c i represents the cost of the i-th design parameter, c i represents the i-th design variable, and C max represents the cost upper limit; Constraint condition 5: Vehicle stability after collision Roll angle limit after collision: g9(x) = Roll Angle(x) - Roll Angle max ≤0 (16) Yaw angle limit after collision: g 10 (x) = Yaw Angle(x) - Yaw Angle max ≤0 (17) In Steps 1 - 4, the optimization problem can be summarized as: Minimize f(x)=(f1(x), f2(x), f3(x)) (18) subject to:g i (x)≤0,i=1,2,...,10 (19) Among them, f1(x) represents the degree of occupant injury, f2(x) represents the vehicle body deformation, f3(x) represents the vehicle weight, and g i (x) represents the constraint condition.

6. The automotive side collision optimization design method based on the co-evolution algorithm according to claim 1, characterized in that Step 5 is specifically carried out in the following steps: Step 5.1: Initialize the parameters of vehicle side collisions for individuals, generate two initial populations P1 and P2. Each individual in the population corresponds to a possible solution; Step 5.2: Evaluate the objective function. For the parameter individuals in the population, calculate their objective values under all objective functions, and perform constraint handling to evaluate whether the individuals meet the constraint conditions in the problem. If an individual meets all constraint conditions, add it to the archive. When the archive is full, sort all the feasible solutions found and select the top N optimal individuals to add to the archive; Step 5.3: Population P1 generates offspring using differential evolution, and population P2 generates offspring using genetic algorithms. The two populations update the archive A according to their own populations and the generated offspring respectively. Populations P1 and P2 update the next generation according to their own populations and the offspring respectively; Step 5.4: When the population converges to the unconstrained Pareto front, output the final population P and the archive A. Population P is output by combining P1 and P2 and selecting the top N optimal solutions.

7. The method for optimizing the side collision design of an automobile based on a co-evolution algorithm according to claim 6, wherein In step 5.1, in order to store and save the feasible solutions found during the evolution process, an archive set is constructed, which records the better feasible solutions in the current population and the historical populations. For the initialization of the population size N, the formula is as follows: P = {x1, x2,..., x N} (20) Among them, the decision vector x of each individual i =(x i1 , x i2 ,..., x id ) is generated from the solution space by uniform random sampling: Among them, x ij represents the j-th decision variable of the i-th individual, and is uniformly selected from the interval [L j , U j , where L j and U j are the lower and upper bounds of the decision variable x i ; The archive is used to record the feasible solutions searched during the convergence process to the unconstrained Pareto front, so as to ensure that the algorithm can store and track the optimal solution set in multi-objective optimization. At the initialization, the archive is empty and its size is N.

8. The method for optimizing the design of a vehicle side collision based on a co-evolutionary algorithm according to claim 1, characterized in that Step 6 is specifically implemented according to the following steps: Step 6.1, initialize the parameter individuals. Use the population P output in step 5 as the initial population of the auxiliary population in step 6, and use the output archive A as the initial population of the main population. Step 6.2, merge the main population and the auxiliary population, and use the genetic algorithm to generate offspring. Step 6.3, apply the biased penalty strategy in the main population to punish the infeasible solutions searched, and apply the standard deviation-guided auxiliary population perturbation strategy in the auxiliary population to adjust the search direction of the auxiliary population. The calculation formula of the biased penalty strategy is as follows: Among them, F(x i ) and F(x i )′ respectively represent the objective vectors before and after the correction of the individual x i . is a normalized form of the constraint violation degree, where CV min represents the solution with the minimum constraint violation degree in the current solution set, and CV max represents the solution with the most constraint violations in the current solution set. consists of the minimum value of each objective in the objective vectors within the sub-region associated with F(x i ). The standard deviation-guided auxiliary population perturbation strategy dynamically adjusts the intensity of the perturbation by calculating the standard deviation difference between the main population and the auxiliary population. The specific perturbation calculation formula is as follows: Through formula (24), it is possible to add a perturbation term based on the standard deviation to achieve the purpose of exploring the solution space near the infeasible solution; In the perturbation term, the calculation formula of the perturbation intensity σ is: Among them, N(0,1) represents a normal distribution with a mean of 0 and a variance of 1. The introduction of this distribution ensures the randomness and continuity of the perturbation term, which helps to improve the diversity of the solutions. θ1 represents the standard deviation of the main population, and θ2 represents the standard deviation of the auxiliary population. The specific calculation formula of the θ standard deviation is as follows: where N is the number of populations, and f j，i represents the function value of the i-th individual on the j-th objective, represents the average value of the j-th objective function, and M represents the objective dimension; Step 6.4, the main population and the auxiliary population combine with the offspring, and select N solutions according to the fitness ranking to enter the next generation.

Images