Seat whiplash injury resistance performance optimization method based on test and simulation

Abstract

The invention relates to the technical field of automobile safety, and provides a seat whiplash injury resistance optimization method based on test and simulation, and the method comprises the steps: firstly, through the comparative analysis of finite element simulation and whiplash test results of a seat, identifying main factors influencing whiplash injury, including design variables such as the thickness of a headrest rod and the thickness of a seat backrest supporting tube; sample points required by optimization are collected by using a Harmersley test design method, and an accurate approximation model is established through a radial basis function neural network (RBF). On the basis, a multi-objective genetic algorithm is adopted to carry out multi-objective optimization on seat design variables, and the purpose is to reduce the neck injury index (NIC) and the upper neck tension. The optimized design parameters effectively reduce the neck injury index and the upper neck tension, and the whiplash injury resistance of the seat is further improved. Through simulation verification, the error between the optimization result and the approximation model is controlled within a reasonable range, and the effectiveness and feasibility of the optimization design are verified.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automobile safety technology, in particular to a seat anti-whiplash performance optimization method based on test and simulation. BACKGROUND

[0002] The proportion of rear-end accidents in traffic accidents has risen to 30%, and the casualties caused by rear-end accidents account for 60% of the total damage caused by all traffic accidents. When a car is rear-ended, the impact of the rear car on the front car will cause the front car to suddenly accelerate, and the human body will first lean forward due to inertia, but the neck will lag behind due to the inertia of the head, and will be "pulled" at the moment; then the head will rebound backward, and the neck will again bear the force in the opposite direction, like the swinging process of a whip being swung, this repeated flexion and extension, torsion can easily cause damage to the cervical ligament, muscle and even vertebral body.

[0003] In order to reduce the whiplash injury, whiplash test is introduced to evaluate the protection ability of car seat to the neck in rear-end collision. Each seat manufacturer considers the whiplash test when developing the seat, and there are mainly two research methods for seat whiplash test on the market, namely slide table simulation of rear-end collision and finite element simulation. The slide table test is an actual test using BioRID-II dummy to simulate the driver, and the collision speed is specified by C-NCAP; while the finite element simulation aims to save manpower and material resources by using finite element method to simulate and optimize the safety performance of the seat: some scholars have determined the influence of backrest stiffness, headrest stiffness, headrest position and head clearance on the whiplash test score through multiple rounds of simulation calculation. Some scholars have improved the whiplash score by improving the seat structure based on the analysis of the force on the human head and neck in rear-end collision. Some scholars found that there is a large error between the simulation results and the experimental results, and by adjusting the sitting posture of the dummy, the test state of the seat and the smoothness of the seat foam skin, the consistency of simulation and test is optimized.

[0004] However, the existing technology still faces some challenges in optimizing the whiplash performance of the seat. First of all, the seat is a system composed of various materials (sheet metal, foam, plastic, etc.) and complex structures, and there are many parameters affecting its whiplash performance, and there may be complex nonlinear coupling relationships between the parameters. The traditional "trial and error method" or single variable optimization method is inefficient and difficult to find the global optimal solution. Secondly, the calculation of a high-precision finite element model is time-consuming, and it is difficult to bear the cost of computing resources if it is directly used in the optimization process of multiple iterations. In addition, how to reduce the main injury indicators (such as NIC and upper neck tension) while ensuring that other neck loads do not exceed the safety limit is a typical multi-objective optimization problem that requires efficient methods to support.

[0005] Therefore, there is an urgent need in the art for a test and simulation based seat anti-whiplash performance optimization method that can comprehensively consider multiple key design variables and significantly improve the anti-whiplash performance of the automobile seat under the premise of controlling the cost. SUMMARY

[0006] The present application provides a test and simulation based seat anti-whiplash performance optimization method that can overcome some or all of the deficiencies of the prior art.

[0007] A test and simulation based seat anti-whiplash performance optimization method according to the present application comprises the following steps:

[0008] S1, establishing a finite element model of a seat system, the model comprising a seat skeleton, a headrest, a backrest foam, a cushion foam and a Biorid-II dummy model;

[0009] S2, performing dynamic simulation analysis on the finite element model according to the C-NCAP whiplash test procedure with a specified acceleration waveform as the input load, and obtaining neck injury simulation data;

[0010] S3, comparing the neck injury simulation data with real vehicle whiplash test data to verify the effectiveness of the finite element model;

[0011] S4, identifying and determining multiple key design variables affecting the seat anti-whiplash performance based on the verified finite element model;

[0012] S5, using the Hamersley test design method to collect sample points within the value range of the key design variables, and obtaining the neck injury response value corresponding to each sample point through finite element simulation;

[0013] S6, constructing an approximate model between the key design variables and the neck injury response value using a radial basis function neural network based on the collected sample points and their response values;

[0014] S7, performing accuracy verification on the approximate model to ensure that its determination coefficient is greater than 0.9;

[0015] S8, using a multi-objective genetic algorithm to perform multi-objective optimization on the approximate model to obtain a Pareto optimal solution set, with reducing the neck injury index and the upper neck tension as the optimization objectives and the remaining neck loads as the constraint conditions;

[0016] S9, selecting a group of optimal solutions from the Pareto optimal solution set and rounding the solution values to engineering achievable parameter values;

[0017] S10, substituting the rounded parameter values into the finite element model for simulation verification to confirm that the optimized seat whiplash performance is improved.

[0018] As preferred, in S1, the seat three-dimensional model is imported into Hypermesh for finite element modeling, the headrest, backrest and cushion are divided by solid elements, considering that the foam will be extruded and deformed during the experiment, tetrahedral elements are used to prevent negative volume, the seat skeleton is sheet metal, which can be divided by shell elements; Then connect the seat, pre-press the dummy, place the dummy in the required position, set the corresponding contact and set the appropriate calculation time and step, and submit the calculation with the whipping test waveform as the input load.

[0019] As preferred, in S2, the whipping test is carried out in an environment of 22.5°C and relative humidity of 10%-70%, the seat and the rack are fixed on the moving trolley according to the original vehicle structure, and the BioRID II dummy is placed on the seat to simulate the occupant injury during the collision; The trolley is launched with an acceleration waveform of 15.65 km / h speed change.

[0020] As preferred, in S4, the key design variables include: headrest rod thickness X1, headrest rod reinforcement thickness X2, seat backrest support pipe thickness X3, backrest left side support plate thickness X4, backrest right side support plate thickness X5, angle adjuster stiffness coefficient X6, backrest foam stiffness coefficient X7, headrest foam stiffness coefficient X8;

[0021] Because the seat structure has symmetry, the symmetric structure needs to be optimized synchronously in the European process, so the backrest left side support plate thickness X4 and the backrest right side support plate thickness X5 are designed by Link option binding

[0022] As preferred, the value range of the key design variables is respectively:

[0023] Headrest rod thickness X1: 0.6 mm ~ 1.8 mm;

[0024] Headrest rod reinforcement thickness X2: 1.0 mm ~ 2.0 mm;

[0025] Seat backrest support pipe thickness X3: 0.5 mm ~ 1.5 mm;

[0026] Backrest left side support plate thickness X4: 1.0 mm ~ 2.0 mm;

[0027] Backrest right side support plate thickness X5: 1.0 mm ~ 2.0 mm;

[0028] Angle adjuster stiffness coefficient X6: 0.8 ~ 1.2;

[0029] Backrest foam stiffness coefficient X7: 0.8 ~ 1.2;

[0030] The headrest foam stiffness coefficient X8 is 0.8 ~ 1.2.

[0031] As a preferred option, in S5, the Hammersley sample points are determined by the following formula:

[0032]

[0033] In the formula, A non-negative integer expressed as a polynomial, where X is the number of samples. Represent any set of numbers: express The function is expressed as:

[0034]

[0035] in Representing an interval Integers within, It is an integer greater than 2.

[0036] Preferably, in S6, the radial basis function uses an interpolation model that approximates the estimation of unknown points. express:

[0037]

[0038] In the formula The coefficients of the difference model representing the radial basis functions are obtained from the N+1 unknown coefficients of the sample points using the following linear equation:

[0039]

[0040]

[0041] In the formula It is a set of radial basis functions; is the Euclidean geometric distance between the point to be measured and the sample point; c is the strip shape parameter;

[0042] The coefficient of determination is used to evaluate how close the approximate model is to the actual results:

[0043]

[0044] In the formula, For the sample size, These are actual measured values. This is an estimated value. This is the average of the actual measured values.

[0045] As a preferred option, in S8, a multi-objective mathematical model for the seat is established:

[0046]

[0047] In the formula This is the lower limit for the thickness variation of the design variable. The initial thickness for design variables, This represents the upper limit of thickness variation for the design variables;

[0048] The constraints include: the shear force of the upper neck, the shear force of the lower neck, the tensile force of the lower neck, the torque of the upper neck, and the torque of the lower neck shall not exceed the high-performance values ​​specified by C-NCAP;

[0049] A multi-objective genetic algorithm was used to calculate the mathematical model established by optimizing the objective and constraints. The dummy's neck injury index (NIC) value was plotted on the ordinate, and the upper neck tension on the dummy was plotted on the lateral. Using x as the x-axis, we obtain the optimal solution.

[0050] As a preferred option, the rounded optimal parameter combination described in S9 is:

[0051] Headrest rod thickness: 1.8 mm;

[0052] Headrest rod reinforcement thickness: 1.8 mm;

[0053] Seat back support tube thickness: 1.5 mm;

[0054] Thickness of the left side support plate of the backrest: 1.0 mm;

[0055] Thickness of the right side support plate of the backrest: 1.0 mm;

[0056] Angle adjuster stiffness coefficient: 1.17;

[0057] Backrest foam stiffness coefficient: 0.93;

[0058] Headrest foam stiffness coefficient: 1.14.

[0059] The beneficial effects of this invention are as follows:

[0060] This invention first identifies the main factors affecting whiplash injury through comparative analysis of finite element simulation and whiplash test results of the seat, including design variables such as headrest rod thickness and seat back support tube thickness. Using the Hammersley design of experiments method, sample points required for optimization are collected, and an accurate approximate model is established using a radial basis function neural network (RBF). Based on this, a multi-objective genetic algorithm is employed to optimize the seat design variables in multiple objectives, aiming to reduce the neck injury index (NIC) and upper neck tension.

[0061] Through systematic multi-objective optimization design, this invention can effectively and specifically reduce the two indicators that cause the most severe neck injuries to occupants in rear-end collisions—the Neck Injury Index (NIC) and upper neck tension. The optimized seat model improved its total score in the C-NCAP whiplash test from 3.23 points to 4.12 points, a significant increase, demonstrating a substantial improvement in its whiplash resistance performance.

[0062] This invention creatively combines Hammersley's experimental design, radial basis function neural network approximation model, and multi-objective genetic algorithm. This method can efficiently handle complex nonlinear relationships between multiple design variables with fewer sample points, and uses a high-precision approximation model to replace time-consuming direct finite element calculations. This makes in-depth and comprehensive multi-objective optimization possible under conventional computing resources, greatly improving optimization efficiency and feasibility.

[0063] The optimization process of this invention does not solely pursue the minimization of a single indicator, but rather focuses on reducing the tension in the NIC and upper neck, while simultaneously considering other key injury indicators such as shear force, tension, and torque in the upper and lower neck as constraints. This ensures that the optimization scheme improves the main performance without leading to an unexpected increase in the risk of injury to other parts, guaranteeing the comprehensiveness and reliability of the optimization.

[0064] The design variables selected in this invention (such as the thickness of each structural component and stiffness coefficient) are all measurable and controllable engineering parameters in seat design. The optimization results provide specific and clear parameter rounded values ​​that match the material thickness specifications in industrial production. Error analysis shows that the optimization results are in good agreement with the actual situation. Therefore, this optimization scheme can directly and smoothly guide the design and production improvement of seat products.

[0065] This invention is not only applicable to the specific seats in this embodiment, but can also be extended to the optimization design of whiplash performance or other safety performance (such as frontal collision and side collision) of seats in other vehicle models, and has broad applicability and promotional value. Attached Figure Description

[0066] Figure 1 This is a flowchart illustrating a method for optimizing the whiplash resistance of a seat based on experimentation and simulation, as described in this embodiment.

[0067] Figure 2 This is a schematic diagram of the pre-compressed seat back and seat cushion in the embodiment;

[0068] Figure 3 This is a schematic diagram of the simulation model for the seat whipping test in the embodiment. Detailed Implementation

[0069] To further understand the content of this invention, a detailed description of the invention will be provided in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0070] Example

[0071] like Figure 1 As shown, this embodiment provides a method for optimizing the whiplash injury resistance performance of a car seat based on experiments and simulations, which includes the following steps:

[0072] S1. Establish a finite element model of the seating system, which includes the seat frame, headrest, backrest foam, seat cushion foam, and Biorid-II dummy model.

[0073] In S1, the 3D model of the seat is imported into Hypermesh for finite element modeling. The headrest, backrest, and seat cushion are meshed using solid elements. Considering that the foam will be compressed and deformed during the experiment, tetrahedral elements are used to prevent negative volume. The seat frame is a sheet metal part, which can be meshed using shell elements. Then, the seat is connected, and the dummy is pre-compressed and placed in the required position. The corresponding contact is set, and appropriate calculation time and step size are set. The whiplash test waveform is used as the input load for calculation. The pre-compressed seat backrest and seat cushion are shown below. Figure 2 As shown. The overall model is as follows. Figure 3 As shown.

[0074] S2. According to the C-NCAP whiplash test procedure, the finite element model is subjected to dynamic simulation analysis using the specified acceleration waveform as the input load, and neck injury simulation data is obtained.

[0075] In S2, the whiplash test was conducted in an environment of 22.5℃ and relative humidity of 10%-70%. The seat and platform were fixed on the moving sled according to the original vehicle structure, and the BioRID II dummy was placed on the seat to simulate the occupant injury during the collision. The sled was launched with an acceleration waveform with a speed change of 15.65km / h.

[0076] Table 1 shows the C-NCAP whiplash test scores for the BioRID II dummy. The results indicate that the dummy's NIC score was only 0.35, a score rate of 17.5%, suggesting poor protection against neck injury. Simultaneously, the upper neck tensile force was 509.56 N, with a score of only 1.42, a score rate of 94.6%, indicating a certain risk of injury. Overall, the seat's protection in the upper neck area is unsatisfactory, with the NIC score being the lowest and the main factor limiting overall whiplash resistance improvement. Therefore, future seat optimization designs should focus on reducing the relative acceleration of the dummy's neck and improving the head and neck support structure to effectively reduce the neck injury index and enhance the seat's whiplash resistance.

[0077] Table 1. C-NCAP Whiplash Test Scores

[0078]

[0079] S3. Compare the simulated neck injury data with the actual vehicle whiplash test data to verify the effectiveness of the finite element model.

[0080] The results of the whiplash simulation calculations were imported into HyperView to obtain a simulation analysis animation. The animation was compared with the dummy's posture during the collision, captured by a high-speed camera during the actual test. The high-speed camera showed the dummy's head initially contacting the headrest at 67ms, while the simulation animation showed the head contact time at 71ms. The seat reached maximum deformation at 122ms and then began to rebound, with the simulation animation showing maximum deformation at 118ms. The finite element calculation results were used to plot damage curves in Hyperworks, which were compared with the whiplash test damage curves to verify the accuracy of the finite element model. The comparison showed that the simulated values ​​and experimental values ​​were relatively consistent, indicating a high degree of fit. This finite element model can be further studied and optimized.

[0081] S4. Based on the validated finite element model, identify and determine several key design variables that affect the seat's anti-whiplash performance.

[0082] The experimental and simulation results show that the main areas of failure are the dummy neck injury index (NIC) and upper neck tension. The NIC (Neural Neck Injury Measurement) primarily reflects the combined effect of the horizontal relative acceleration and velocity difference between the dummy's head and chest (i.e., the position of the first thoracic vertebra). The more intense the relative motion between the dummy's head and chest, the greater the flexion-extension deformation of the cervical spine, thus leading to a higher risk of neck injury.

[0083] The relative acceleration is:

[0084]

[0085] The relative speed is:

[0086]

[0087] In the formula, The horizontal acceleration of the dummy's head relative to its chest; The horizontal velocity of the dummy's head relative to its chest; The horizontal acceleration of the chest; This refers to the horizontal acceleration of the head.

[0088] As can be seen from the formula, NIC is obtained by integrating the horizontal acceleration of the dummy's head relative to its chest. It mainly examines the relative motion trend between the dummy's head and chest. Therefore, the smaller the acceleration difference between the dummy's head and chest, that is, the more synchronized the relative motion between the head and chest, the lower the value of NIC will be.

[0089] Analysis of the seat structure and human body structure reveals that eight factors significantly impact the whiplash performance of the seat: headrest rod thickness, headrest rod reinforcement thickness, seat back support tube thickness, left back support plate thickness, right back support plate thickness, adjuster stiffness coefficient, back foam stiffness coefficient, and headrest foam stiffness coefficient. These eight factors are selected as design variables. Due to the symmetry of some structural components in the seat, these symmetrical components need to be optimized simultaneously during the Europeanization process. Therefore, the thicknesses of the left and right back support plates are designed using the Link option, with the value range shown in Table 2.

[0090] Table 2 Range of Design Variables

[0091]

[0092] S5. Using the Hammersley experimental design method, sample points are collected within the range of values ​​of the key design variables, and the neck injury response value corresponding to each sample point is obtained through finite element simulation.

[0093] Hammersley sampling utilizes a pseudo-random numerical generator to uniformly sample within a hypercube. It is suitable not only for highly nonlinear response surfaces but also, when there are many design variables and a wide sampling space, can output reliable results with relatively few sample points. Compared to Latin hypercube sampling, it offers better uniformity. Therefore, this embodiment employs Hammersley sampling to design 82 experiments and sample 8 sets of optimization variables. The Hammersley sample points are determined by the following formula:

[0094]

[0095] In the formula, A non-negative integer expressed as a polynomial, where X is the number of samples. Represent any set of numbers: express The function is expressed as:

[0096]

[0097] in Representing an interval Integers within, It is an integer greater than 2.

[0098] S6. Based on the collected sample points and their response values, construct an approximate model between the key design variables and the neck injury response values ​​using a radial basis function neural network.

[0099] Radial basis function (RBF) neural network models, compared to least squares and moving least squares methods, are less robust and less prone to overfitting when the model complexity is high, thus better meeting the accuracy requirements of approximate models. Because the RBF method has a good ability to approximate highly complex functions, its results exhibit good nonlinear approximation, and it has the advantages of fast convergence and computational stability, this embodiment uses the RBF to establish the approximate model. The RBF is used as an interpolation model to approximate the unknown points. express:

[0100]

[0101] In the formula The coefficients of the difference model representing the radial basis functions are obtained from the N+1 unknown coefficients of the sample points using the following linear equation:

[0102]

[0103]

[0104] In the formula It is a set of radial basis functions; is the Euclidean geometric distance between the point to be measured and the sample point; c is the strip shape parameter;

[0105] The coefficient of determination is used to evaluate how close the approximate model is to the actual results:

[0106]

[0107] In the formula, For the sample size, These are actual measured values. This is an estimated value. This is the average of the actual measured values.

[0108] S7. Verify the accuracy of the approximate model to ensure that its coefficient of determination is greater than 0.9.

[0109] The approximate model accuracy of the response model is greater than 0.9, indicating that the calculated value fits the true value better and can be used to replace the actual model for subsequent optimization.

[0110] S8. Taking the reduction of neck injury index and upper neck tension as optimization objectives, and the remaining neck loads as constraints, a multi-objective genetic algorithm is used to perform multi-objective optimization on the approximate model to obtain the Pareto optimal solution set.

[0111] An approximate model based on the RBF radial basis function was established, and a multi-objective genetic algorithm (MOGA) was used to optimize the seat design variables. In this paper's seat whiplash test, the dummy's neck injury value (NIC) and upper neck tension were the most frequently lost scores. Therefore, the dummy's neck injury value (NIC) and upper neck tension were... As the optimization objective, the shear force of the upper neck, the shear force of the lower neck, the tensile force of the lower neck, the torque of the upper neck, and the torque of the lower neck are used as constraints. A multi-objective mathematical model of the seat is established:

[0112]

[0113] In the formula This is the lower limit for the thickness variation of the design variable. The initial thickness for design variables, The upper limit of the thickness variation of the design variables is set. A multi-objective genetic algorithm is used to calculate the mathematical model established based on the above optimization objectives and constraints. The dummy's neck injury index (NIC) value is plotted on the ordinate, and the upper neck tension of the dummy is plotted on the lateral. Using x as the x-axis, we obtain the Pareto optimal solution set.

[0114] S9. Select a set of optimal solutions from the Pareto optimal solution set and round the solution values ​​to the parameter values ​​that can be achieved in the project.

[0115] As the NIC index increases, the tension in the upper neck decreases continuously, showing a decreasing trend. Therefore, it is necessary to select the optimization results from the optimal solution set based on the scores of various indicators in C-NCAP. The rounded results of the optimized parameters are shown in Table 3.

[0116] Table 3. Optimization results of design variables for thickness

[0117]

[0118] S10. Substitute the rounded parameter values ​​into the finite element model for simulation verification to confirm that the optimized seat whiplash performance has been improved.

[0119] The optimized part dimensions were reintroduced into the finite element model, and the whipping test simulation was performed again. The calculation results were compared with the results obtained from the optimization of the approximate model to verify the feasibility of the optimization results.

[0120] The optimized simulation results show some error compared to the approximate model. This is because the actual production thickness of parts differs from that in the approximate model. However, the error is below 7%, indicating that the proxy model has high accuracy, thus the optimized results are valuable for reference. The optimized simulation results were then re-scored according to C-NCAP and compared with the scores before optimization. Before optimization, the whiplash simulation test score for the seat model was 3.23 points, with a maximum NIC value of 26.14 and a score of 0.35 points. The upper neck tension... The maximum value was 509.56 N, with a score of 1.42. All other indicators were good. After optimization, the maximum NIC value decreased to 17.71, and the score improved to 1.12; upper neck tension... The maximum value decreased to 404.38, and the score increased by 1.5 points. No points were lost in any of the other indicators, and performance in many indicators improved compared to before optimization. The whipping score increased from 3.23 points to 4.12 points, as shown in Table 4.

[0121] Table 4 Comparison of Whiplash Scores Before and After Optimization

[0122]

[0123] This embodiment analyzes a whiplash test on a front seat to explore its protective performance for the human neck in a rear-end collision and proposes an optimization scheme. The effectiveness of the model is verified through a combination of finite element simulation and whiplash testing, providing a reliable basis for improving the safety of seat design.

[0124] First, by comparing and analyzing the results of finite element simulation and whiplash tests of the seat, the main factors affecting whiplash injury were identified, including design variables such as the thickness of the headrest rod and the thickness of the seat back support tube. Using the Hammersley design of experiments method, sample points required for optimization were collected, and an accurate approximate model was established using a radial basis function neural network (RBF). Based on this, a multi-objective genetic algorithm was used to optimize the seat design variables, aiming to reduce the neck injury index (NIC) and upper neck tension. After optimization, the seat whiplash score improved from 3.23 to 4.12. The optimized design parameters effectively reduced the NIC and upper neck tension, further improving the seat's resistance to whiplash injury. Simulation verification showed that the error between the optimized results and the approximate model was controlled within a reasonable range, validating the effectiveness and feasibility of the optimized design.

[0125] This embodiment successfully established an optimized design process for seat whiplash performance. By combining simulation and experimentation, it not only improved seat safety but also provided a reliable optimization solution for seat design. Future research can further consider other safety performance indicators and explore more optimization methods to achieve a more comprehensive improvement in seat safety.

[0126] The present invention and its embodiments have been described above illustratively. This description is not restrictive, and the figures shown are only one embodiment of the present invention; the actual structure is not limited thereto. Therefore, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the present invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for optimizing seat whiplash performance based on testing and simulation, characterized by: The method comprises the following steps: S1, a finite element model of a seat system is established, the model comprising a seat framework, a headrest, backrest foam, a cushion foam and a Biorid-II dummy model; S2, according to the C-NCAP whipping test procedure, a specified acceleration waveform is taken as an input load to perform dynamic simulation analysis on the finite element model, and neck injury simulation data is obtained; S3, the neck injury simulation data is compared with real vehicle whipping test data to verify the effectiveness of the finite element model; S4, based on the verified finite element model, a plurality of key design variables affecting the seat anti-whipping performance are identified and determined; S5, a Hammersley test design method is used to collect sample points within the value range of the key design variables, and the neck injury response values corresponding to each sample point are obtained through finite element simulation; S6, based on the collected sample points and their response values, a radial basis function neural network is used to construct an approximate model between the key design variables and the neck injury response values; S7, the accuracy of the approximate model is verified to ensure that the determination coefficient is greater than 0.9; S8, taking reduction of the neck injury index and the upper neck tension as optimization objectives, and taking the remaining neck load as a constraint condition, a multi-objective genetic algorithm is used to perform multi-objective optimization on the approximate model to obtain a Pareto optimal solution set; S9, a group of optimal solutions is selected from the Pareto optimal solution set, and the solution values are rounded to engineering achievable parameter values; S10, the rounded parameter values are substituted into the finite element model for simulation verification to confirm that the optimized seat whipping performance is improved.

2. The method of claim 1, wherein: In S1, the seat three-dimensional model is imported into Hypermesh for finite element modeling, the headrest, backrest and cushion are divided by solid elements, considering that the foam will be extruded and deformed during the experiment, tetrahedral elements are used to prevent negative volume, the seat framework is a sheet metal part, which can be divided by shell elements; then the seat is connected, the dummy is pre-pressed, the dummy is placed in the required position, the corresponding contact is set and the appropriate calculation time and step are set, and the whipping test waveform is taken as the input load to submit calculation.

3. The method of claim 2, wherein: In S2, the whipping test is performed in an environment of 22.5℃ and relative humidity of 10%-70%, the seat and the rack are fixed on the moving trolley according to the original vehicle structure, and the BioRID II dummy is placed on the seat to simulate the occupant injury during the collision process; the trolley is launched with an acceleration waveform with a speed change of 15.65km / h.

4. The method of claim 3, wherein: In S4, the key design variables include: headrest rod thickness X1, headrest rod reinforcement thickness X2, seat backrest support pipe thickness X3, backrest left side support plate thickness X4, backrest right side support plate thickness X5, angle adjuster stiffness coefficient X6, backrest foam stiffness coefficient X7, headrest foam stiffness coefficient X8; Because the seat part structure has symmetry, the symmetric structure needs to be optimized synchronously in the European process, therefore, the backrest left side support plate thickness X4 and the backrest right side support plate thickness X5 are designed by Link option binding.

5. The method of claim 4, wherein: The value ranges of the key design variables are respectively: Headrest rod thickness X1: 0.6 mm ~ 1.8 mm; Headrest rod reinforcement thickness X2: 1.0 mm ~ 2.0 mm; Seat back support tube thickness X3: 0.5 mm ~ 1.5 mm; Back left side support plate thickness X4: 1.0 mm ~ 2.0 mm; Back right side support plate thickness X5: 1.0 mm ~ 2.0 mm; Rocker rigidity coefficient X6: 0.8 ~ 1.2; Back foam rigidity coefficient X7: 0.8 ~ 1.2; Headrest foam rigidity coefficient X8: 0.8 ~ 1.

2.

6. The method of claim 5, wherein: In S5, the Hammerstein sample point is determined by the following formula: ； wherein is a non-negative integer represented by a polynomial, X is the number of samples any set of sequences represented by is represented by a function of ； wherein represents an integer within the interval , is an integer greater than 2.

7. The method of claim 6, wherein: In S6, the radial basis function is used to approximate the interpolation model of the unknown point Indicates: ； wherein denote the difference model coefficients of the radial basis functions, the sample points are obtained from the following linear equation for N + 1 unknown coefficients: ； ； wherein is a set of radial basis functions; is the Euclidean geometric distance between the point to be measured and the sample point; c is a bar shape parameter; The determination coefficient is used to evaluate the closeness of the approximation model to the true result: ； wherein is the number of samples, is the actual measured value, is the estimated value, is the average of the actual measured values.

8. The method of claim 7, wherein: In S8, a multi-objective mathematical model of the seat is established: ； wherein a lower limit of thickness variation as a design variable, an initial thickness as a design variable, an upper limit of thickness variation as a design variable; The constraint conditions include: the upper neck shear force, the lower neck shear force, the lower neck tension, the upper neck torque, and the lower neck torque are all not exceeding the high performance value specified by C-NCAP; A multi-objective genetic algorithm was used to calculate the mathematical model established by optimizing the objective and constraints. The dummy neck injury index (NIC) value was plotted on the ordinate, and the upper neck tension of the dummy was plotted on the lateral. Using x as the x-axis, we obtain the optimal solution.

9. The method of claim 8, wherein: The rounded optimal parameter combination in S9 is: Headrest rod thickness: 1.8 mm; Headrest rod reinforcement thickness: 1.8 mm; Seat back support tube thickness: 1.5 mm; Back left side support plate thickness: 1.0 mm; Back right side support plate thickness: 1.0 mm; Rocker rigidity coefficient: 1.17; Back foam rigidity coefficient: 0.93; Headrest foam rigidity coefficient: 1.14.

Images