Vehicle virtual evaluation method and system based on centroid and lyapunov index

Abstract

The present application belongs to the field of vehicle testing, and a vehicle virtual evaluation method based on centroid and Lyapunov index comprises: S1: obtaining centroid motion data of an occupant in a vehicle collision process and kinematics response data of multiple key parts of the occupant; S2: based on the centroid motion data, calculating an extrapolated centroid index for evaluating global dynamic stability of the occupant and a maximum Lyapunov index for evaluating local dynamic stability of the occupant; based on the kinematics response data of the multiple key parts of the occupant, calculating multiple local injury indexes of the occupant respectively; S3: based on the extrapolated centroid index, the maximum Lyapunov index and the multiple local injury indexes, classifying the injury mode of the occupant according to a preset injury mode classification rule. The present application solves the problem that the existing vehicle virtual evaluation technology only relies on local physical indexes and cannot quantitatively evaluate the overall dynamic stability and injury mode of the occupant, resulting in that the safety risk identification is not comprehensive and the design optimization direction is not clear.

Description

Technical Field

[0001] This invention belongs to the field of vehicle testing, and in particular relates to a virtual vehicle evaluation method and system based on the center of mass and Lyapunov index. Background Technology

[0002] Virtual collision safety assessment is a core technology in the modern automotive safety development process. By simulating the physical process of a collision with high precision in a computer environment, it is possible to predict and optimize the protection performance of the vehicle structure and restraint system for occupants before the physical prototype is manufactured. This can significantly shorten the development cycle, reduce development costs, and systematically improve the passive safety level of the vehicle.

[0003] Currently, virtual assessment systems generally adopt an evaluation paradigm based on local physical response. This method relies on high-precision human biomechanical dummy models to monitor the physical response values ​​of specific anatomical locations (such as the head, chest, neck, and thighs) during a collision and compare them with preset biomechanical injury thresholds to determine safety. Typical assessment indicators include the Head Injury Criterion (HIC), chest compression, and neck force and torque. Although this method is clear and mature in assessing local damage, it has a fundamental flaw: First, the assessment indicators focus on the peak response of specific anatomical locations at a certain moment or under quasi-static conditions (such as maximum chest compression), failing to characterize the overall motion trend of the occupant as a multibody dynamic system throughout the complete collision sequence. For example, during a collision, the occupant's center of gravity may shift significantly due to the impact. Although this may not immediately cause local indicators to exceed the limits, it foreshadows that the occupant will leave the effective restraint area of ​​the seatbelt and engage in a secondary collision with the vehicle's interior components. This assessment method has a safety blind spot, failing to identify potentially high-risk scenarios caused by overall loss of control of motion, thus significantly reducing the predictive accuracy and completeness of virtual assessments.

[0004] Second, during a collision, the dynamic stability of the occupant's body is a key factor determining the final degree of injury. If the restraint system can maintain the occupant's relatively stable posture, even if a certain amount of impact is absorbed locally, the overall damage can be controlled. Conversely, if the occupant's posture becomes uncontrollable (such as during rotation or projectile impact), it may trigger a chain reaction of injuries to multiple parts of the body. Existing methods only provide scalar quantities of the damage results and lack quantitative means for assessing the dynamic attribute of motion stability. This makes it impossible to effectively distinguish between two fundamentally different safety scenarios: localized damage with overall controllability and localized damage meeting safety standards but overall loss of control. Consequently, the evaluation conclusions lack discriminatory power, making it difficult to guide engineers in specifically optimizing restraint systems to improve their stability control performance, thus reducing the efficiency and specificity of optimization design.

[0005] Third, different collision conditions and different failure modes of constraint systems can induce occupant motion patterns with very different characteristics. During the evaluation, only isolated damage values ​​are output, and it is impossible to correlate these values ​​with the underlying kinematic causes. This leads to unclear directions for engineering optimization. For example, when faced with high head injury values, engineers find it difficult to determine whether the root cause is excessive forward tilting or lateral swaying. As a result, they cannot optimize seat belts, front airbags, or side airbag systems in a targeted manner. This lack of diagnostic capability prolongs the trial and error cycle of design iteration, increases development costs, and also restricts the in-depth optimization of safety performance. Summary of the Invention

[0006] This invention provides a vehicle virtual evaluation method and system based on the center of mass and Lyapunov index, which solves the problem that existing vehicle virtual evaluation technologies rely only on local physical indicators and cannot quantitatively evaluate the overall dynamic stability and damage mode of occupants, resulting in incomplete safety risk identification and unclear design optimization direction.

[0007] This invention provides a basic solution: a virtual vehicle evaluation method based on the center of mass and Lyapunov index, comprising the following steps: S1: Acquire the center of mass motion data of the occupant and the kinematic response data of multiple key parts of the occupant during the vehicle collision. The center of mass motion data includes the center of mass position and center of mass velocity of the occupant. S2: Based on the center of mass motion data, calculate the extrapolated center of mass index for assessing the global dynamic stability of the occupant and the maximum Lyapunov index for assessing the local dynamic stability of the occupant; calculate multiple local injury indices of the occupant based on the kinematic response data of multiple key parts of the occupant. S3: Based on the extrapolated centroid index, the maximum Lyapunov index, and multiple local damage indices, the damage patterns of the occupants are classified according to the preset damage pattern classification rules.

[0008] Preferably, in step S2, the extrapolation centroid index is calculated using the following formula:

[0009] in, The location of the center of mass. Let be the velocity of the center of mass, h be the height of the center of mass, and g be the acceleration due to gravity.

[0010] More preferably, in step S2, the calculation steps for the maximum Lyapunov exponent are as follows: 1) Preprocess and fuse the center-of-mass motion data to construct a fused time series containing position and velocity information; 2) Based on the fused time series, the system dynamics phase space is constructed using the phase space reconstruction method; 3) In phase space, calculate the average exponential divergence rate of neighboring trajectory points over time, and determine the fitting slope of the average exponential divergence rate as the maximum Lyapunov exponent.

[0011] More preferably, in the system dynamics phase space, the delay time τ is determined by the CC method, the embedding dimension m is determined by the spurious nearest neighbor method, and the fused time series is reconstructed into an m-dimensional phase space trajectory based on the τ and m.

[0012] Preferably, determining the maximum Lyapunov index specifically includes: In the reconstructed phase space, find the nearest neighbor point that satisfies the distance threshold for each trajectory point to form multiple neighbor pairs; Track the separation distance of each neighboring point pair over the evolution time step, and calculate the average log distance of all point pairs at each evolution step; A linear fit was performed on the curve of the mean logarithmic distance as a function of the evolution step size, and the slope of the fitted line was used as the maximum Lyapunov exponent.

[0013] Preferably, in step S1, the key areas include the head, neck, chest, and thigh; in step S2, the local injury indicators include head injury criteria, neck injury indicators, chest compression, and femoral axial force.

[0014] Preferably, in step S3, the damage mode classification rule is as follows: The occupant injury mode is determined to be a Class A stable and controllable mode when all three of the following conditions are met simultaneously. a) The trajectory of the extrapolated centroid index lies within the boundary of a predefined virtual support base region throughout the entire collision process; b) The maximum Lyapunov exponent is less than the first stability threshold; c) All local damage indicators are below the preset excellent performance threshold; Otherwise, the occupant injury mode is determined to be an unstable mode.

[0015] More preferably, the unstable modes include type B rotational runaway mode, type C translational projectile mode, and type D mixed instability; wherein, If the trajectory of the extrapolated centroid index exhibits a rotational characteristic around the vertical axis, and the maximum Lyapunov exponent is greater than the first stability threshold and less than or equal to the second stability threshold, it is determined to be a type B rotational runaway mode. If the trajectory of the extrapolated centroid index exceeds the boundary of the virtual support base region within a preset time window, and the maximum Lyapunov exponent is greater than the second stability threshold, it is determined to be a Class C translational projectile mode. Otherwise, it belongs to type D, a mixed unstable type.

[0016] More preferably, the virtual support base area is defined by the spatial range formed by the vehicle seat, seat belt constraint boundary, and in-vehicle occupant restraint device.

[0017] Another basic solution provided by this invention: a virtual vehicle evaluation system based on the center of mass and Lyapunov index, used to run the above method, including: The data input module is used to acquire the center of mass motion data of the occupants and the kinematic response data of multiple key parts of the occupants during the vehicle collision. The calculation and analysis module is used to calculate the extrapolated center of mass index and the maximum Lyapunov index based on the center of mass motion data, and to calculate multiple local damage indices based on the kinematic response data of the key parts. The damage mode classification module is used to classify the occupant's damage mode based on the extrapolated centroid index, the maximum Lyapunov index, and multiple local damage indices, according to preset damage mode classification rules.

[0018] The principles and advantages of this invention are as follows: 1. By introducing the extrapolated center of mass index and the maximum Lyapunov exponent, the dynamic stability of occupants during the collision process is quantified from both global and local dimensions in the transient analysis of vehicle collisions. This enables dynamic stability tracking and motion trend prediction throughout the collision process, identifies motion runaway risks that traditional local damage indicators cannot capture, improves the predictive completeness and safety assessment depth of virtual evaluation, and can provide early warning of runaway trends during the simulation stage. This allows for optimization of occupant restraint strategies from the source and enhances the foresight of vehicle passive safety development.

[0019] 2. By synergistically analyzing and classifying stability indicators (extrapolated center of mass, maximum Lyapunov index) with traditional local injury indicators (such as HIC, chest compression), different injury patterns can be identified, and the injury results can be directly linked to the deep kinematic causes. Based on the classification results, the root cause of the injury can be quickly determined to be specific motion patterns such as forward tilting, lateral swinging, and rotational projectile. This allows for targeted optimization of restraint systems such as seat belts and airbags, shortening the design iteration cycle, reducing trial and error costs, and improving the efficiency of passive safety performance optimization. Attached Figure Description

[0020] Figure 1 This is a flowchart of the present invention; Figure 2 This is a flowchart illustrating the damage mode classification of the present invention. Detailed Implementation

[0021] The following detailed description illustrates the specific implementation method: The specific implementation process is as follows: Example 1 See Figures 1 to 2A virtual vehicle evaluation method based on centroid and Lyapunov index includes the following steps: S1: Acquire the center of mass motion data of the occupant and the kinematic response data of multiple key parts of the occupant during the vehicle collision. The center of mass motion data includes the center of mass position and center of mass velocity of the occupant. In step S1, the key parts include the head, neck, chest and thigh.

[0022] S2: Based on the center of mass motion data, calculate the extrapolated center of mass index for assessing the global dynamic stability of the occupant and the maximum Lyapunov index for assessing the local dynamic stability of the occupant; calculate multiple local injury indices of the occupant based on the kinematic response data of multiple key parts of the occupant. In step S2, the extrapolated centroid index is calculated using the following formula:

[0023] in, The location of the center of mass. Let be the velocity of the center of mass, h be the height of the center of mass, g be the acceleration due to gravity, and XcoM be the predicted position of the center of mass in the next motion considering gravity. Global stability is quantified by analyzing the relative relationship between the XcoM trajectory and a predefined virtual support base region. The virtual support base region is defined by the spatial range formed by the vehicle seat, seatbelt constraint boundaries, and occupant restraint devices (front airbags / instrument panel). If the extrapolated center of mass index is within the virtual support base region, the human body can maintain balance; if it exceeds the virtual support base region, balance will be lost.

[0024] The boundary parameters of the virtual support base area include the longitudinal boundary (vehicle driving direction, x-axis), the lateral boundary (vehicle width direction, y-axis), and the vertical boundary (vehicle height direction, z-axis). Among them, the longitudinal boundary (vehicle driving direction, x-axis): the front end is the maximum longitudinal coverage position after the airbag is fully deployed (determined by the airbag inflation model and deployment time), and the rear end is the junction position of the seat back and seat cushion (determined by the geometric coordinates of the seat three-dimensional model). Lateral boundary (vehicle width direction, y-axis): the left side is the inner side of the left side panel of the seat, the right side is the inner side of the right side panel of the seat, and if there is a side airbag, it extends to the inner boundary after the side airbag is deployed. Vertical boundary (vehicle height direction, z-axis): the lower end is the position of the upper surface of the seat cushion, and the upper end is the lower surface of the headliner or the maximum vertical height after the airbag is fully deployed (take the smaller of the two values).

[0025] The boundary coordinates of the virtual support base area are extracted from the geometric information of the vehicle finite element model. For example, the typical range of the virtual support base area on the driver's side of a car is: x∈[0.6m,1.5m], y∈[-0.3m,0.3m], z∈[0.4m,1.2m] (with the center of the front wheel of the vehicle as the origin of the coordinate system).

[0026] In step S2, the calculation steps for the maximum Lyapunov exponent are as follows: 1) Preprocess and fuse the center-of-mass motion data to construct a fused time series containing position and velocity information; Wavelet thresholding is used to denoise the time series of centroid position and velocity, and the denoised position and velocity series in the same direction are merged into a two-dimensional time series.

[0027] Specifically, the center-of-mass motion data includes the occupant's center-of-mass position (three-dimensional position coordinates CoM_x(t), CoM_y(t), CoM_z(t)) and center-of-mass velocity (three-dimensional velocity vectors vCoM_x(t), vCoM_y(t), vCoM_z(t)). Wavelet thresholding is applied to the collected CoM and vCoM time series data (taking the longitudinal direction as an example, i.e., CoM_x(t) and vCoM_x(t)) to eliminate high-frequency noise in the simulation data. The db4 wavelet basis function is selected, and the decomposition layer is 5 levels to eliminate high-frequency noise in the simulation data. CoM_x(t) and vCoM_x(t) are then fused into a two-dimensional time series S(t) = [CoM_x(t), vCoM_x(t)], taking into account both position and velocity information to improve the comprehensiveness of the stability analysis.

[0028] 2) Based on the fused time series, the system dynamics phase space is constructed using the phase space reconstruction method; In the system dynamics phase space, the delay time τ is determined by the CC method, the embedding dimension m is determined by the spurious nearest neighbor method, and the fused time series is reconstructed into an m-dimensional phase space trajectory based on the delay time τ and the embedding dimension m.

[0029] Specifically, for the preprocessed time series, the correlation integral is calculated using the CC method, and the delay time τ is determined by finding the first minimum point of the correlation integral curve. The false nearest neighbor method is used to gradually increase the embedding dimension, and the embedding dimension m is determined when the proportion of false nearest neighbors drops below 5%. Based on the delay time τ and the embedding dimension m, the two-dimensional time series S(t) is reconstructed into an m-dimensional phase space vector Y(t), with the formula Y(t) = [S(t), S(t+τ), S(t+2τ), ..., S(t+(m-1)τ)], where t = 1, 2, ..., N-(m-1)τ, and N is the total number of data points. In this embodiment, the delay time τ is set to 5-10 (corresponding to a time interval of 5-10ms when the sampling frequency is 1000Hz), and m typically takes a value of 3-5 in collision scenarios.

[0030] 3) In phase space, calculate the average exponential divergence rate of neighboring trajectory points over time, and determine the fitting slope of the average exponential divergence rate as the maximum Lyapunov exponent.

[0031] The following methods can be used to determine the maximum Lyapunov exponent: a. In the reconstructed phase space, select an initial point and determine its nearest neighbor that satisfies a distance threshold condition to obtain a pair of neighboring points, wherein the distance threshold condition is: the Euclidean distance between the nearest neighbor and the initial point is less than a preset threshold ε1; calculate the initial distance between the pair of neighboring points and track the separation distance of the pair over time; perform a linear fit on the relationship between the logarithm of the separation distance and the evolution time, and determine the slope of the fitted line as the maximum Lyapunov exponent.

[0032] Specifically, in the reconstructed phase space, an initial point X(t0) is selected, and the point with the smallest Euclidean distance in the entire phase space is found as its nearest neighbor X(t1), satisfying ||Y(t)-Y(t0)||<ε1, where ε1 is the nearest neighbor threshold, taken as 0.1-0.2 times the standard deviation of the distances between all points in the phase space; the Euclidean distance d0 between the initial point X(t0) and its nearest neighbor X(t1) is calculated, where d0=||X(t0)||. X(t1)||; and over time (evolution time step t), track the separation distance d(t) between the initial point X(t0) and its nearest neighbor X(t1), where d(t) = ||X(t+t0)||. X(t+t1)||; A linear fit is performed on the sequence ln(d(t)) and the evolution time t. The slope of the fitted line is the maximum Lyapunov exponent λmax, and the fitting relationship is ln(d(t))=λmax. t+C, where C is a constant.

[0033] b. In phase space, find the nearest neighbor points that satisfy the distance threshold condition for multiple reference points to form multiple neighbor pairs. When finding the nearest neighbor point for each reference point, the Euclidean distance between the two points must be less than a preset threshold ε2. For each evolution time step, calculate the logarithm of the ratio of the separation distances of all neighbor pairs and take the average value to obtain a series of average logarithmic distances. Perform linear fitting on the relationship between the average logarithmic distance and the evolution time step, and determine the slope of the fitted line as the maximum Lyapunov exponent.

[0034] Specifically, multiple reference points are selected, and the nearest neighbor vector Y(t0) with the smallest Euclidean distance for each reference point is found in the entire phase space, satisfying ||Y(t) - Y(t0)|| < ε2, where ε2 is the nearest neighbor threshold, typically taken as 0.1 to 0.2 times the data standard deviation. This forms multiple nearest neighbor pairs (Y(t), Y(t0)). For each time interval Δt (typically 1-5 ms), for each valid nearest neighbor pair, the logarithm of the distance ratio between the vectors corresponding to time t and t0 after Δt is calculated, i.e., the trajectory divergence rate ln( ||Y(t+Δt) - Y(t0+Δt)|| / ||Y(t) - Y(t0)|| is calculated. Then, the average of the divergence rates calculated for all point pairs under the same Δt is taken to obtain the average divergence rate under that evolution step size. For different evolution step sizes Δt... A linear fit is performed on the sequence of average divergence rates and the corresponding linear regression line. The slope of the fitted line is the maximum Lyapunov exponent λ_max, where the fitting relationship can be expressed as lnε = λ_max. t + C (where ε² is the average distance scale and C is a constant), or equivalently, the average divergence rate.<ln(d(Δt) / d(0))> ≈ λ_max Δt.

[0035] Regarding phase space reconstruction, a 6-dimensional phase space (x, y, z, vx, vy, vz) is constructed based on the center of mass position and velocity in step S1. In this 6-dimensional phase space, considering computational complexity and data dimensionality, we can focus only on the stability of the main motion direction (e.g., the vehicle's forward direction X), thus using a 2-dimensional phase space (x, vx). The 6-dimensional phase space (x, y, z, vx, vy, vz) is a complete global dynamic description, treating the occupant's center of mass motion as an object moving freely in three-dimensional physical space and three-dimensional velocity space. Motion in any direction (including complex oblique and rotational trends) will be captured. The 2-dimensional phase space (x, vx) is a dimensionality reduction model for the main risks. It assumes that in vehicle collisions (especially frontal and rear-end collisions), the motion along the vehicle's longitudinal axis (X direction) is dominant and the most dangerous. Therefore, it ignores the vertical (Z) and lateral (Y) motions, which have a smaller impact on stability.

[0036] In the initial stage of parameter optimization for vehicle restraint systems (seat belts, airbags), it may be necessary to run simulations thousands of times. Two-dimensional phase space analysis can quickly screen out those design schemes that perform poorly in the main motion direction from a large number of design schemes.

[0037] Next, a detailed calculation procedure of Rosenstein's algorithm based on two-dimensional phase space (a simple algorithm suitable for large datasets) is given: Suppose we have data on the position and velocity of the centroid in the X direction at N time points, namely t(i), x(i), vx(i) (i = 1, 2,..., N); Then the phase space points are: P(i) = [x(i), vx(i)], a total of N points; Parameter settings: Set a minimum time interval (min_t) to avoid selecting adjacent points (because adjacent points are naturally close in phase space), Set a maximum time interval (max_t) for tracking evolution, Set a neighborhood radius (epsilon) to find the nearest neighbor points (optional); For each reference point P(i) (from i = 1 to N - min_t): a. Find the nearest neighbor point P(j) that satisfies the following conditions: Condition 1: |i - j| > min_t (to avoid temporal correlation) Condition 2: ||P(i) - P(j)|| < epsilon (optional, but can be omitted if the data is dense) b. If such a neighboring point is found, record this pair of points (P(i), P(j)), and record the initial distance d0(i) = ||P(i) - P(j)||; For each pair of points (P(i), P(j)), we track their evolution with the time step k, calculate the distance d_k(i) = ||P(i + k) - P(j + k)|| until k reaches max_t or exceeds the data range; For each k (from 1 to max_t), calculate the average value of ln(d_k(i)) for all pairs, that is: S(k) = (1 / M) sum_{i = 1}^{M}ln(d_k(i)) where M is the number of valid pairs; Then, in the curve of S(k) changing with the time step k, the slope of the linear part is the maximum Lyapunov exponent λmax.

[0038] In step S2, the local injury indicators include head injury criteria, neck injury indicators, chest compression, and femoral axial force.

[0039] This example is based on a 100% rigid barrier frontal collision scenario (initial collision speed 50 km / h, front driver's seat, seatbelt pretensioning and front airbag function normal). The virtual support base area boundaries are set as x∈[0.6m,1.5m], y∈[-0.3m,0.3m], z∈[0.4m,1.2m] (with the center of the front wheels of the vehicle as the origin). In this example, the extrapolated centroid index XcoM is calculated from the collected data, and the three axial components are: longitudinal x∈[0.72m,1.41m], lateral y∈[-0.21m,0.18m], and vertical z∈[0.48m,1.12m], with a maximum Lyapunov exponent of 0.21. Among the local damage indices, the head injury criterion is 382 (HIC). 15 The neck injury index was 0.22 (N). ij The chest compression was 22mm, and the axial force on the femur was 3.7kN.

[0040] S3: Based on the extrapolated centroid index, the maximum Lyapunov index, and multiple local damage indices, the damage patterns of the occupants are classified according to the preset damage pattern classification rules.

[0041] In step S3, the damage mode classification rule is as follows: The occupant injury mode is determined to be a Class A stable and controllable mode when all three of the following conditions are met simultaneously. a) The trajectory of the extrapolated centroid index is within the boundary of a predefined virtual support base region throughout the entire collision process; throughout the entire collision process, the three axial components of the extrapolated centroid index XcoM are all within the boundary range corresponding to the virtual support base region. b) The maximum Lyapunov exponent is less than the first stability threshold; the first stability threshold is 0.3 (dimensionless, a stability threshold calibrated based on 100 sets of collision simulation data). c) All local injury indicators are below the preset excellent performance thresholds, which include the excellent performance thresholds for the head, neck, chest, and thigh. In this embodiment, specifically, the excellent performance threshold for the head is 450~550 (HIC). 15 The excellent performance threshold for the neck is 0.25~0.4 (N). ij Excellent performance threshold for the chest is 25~35mm (compression), and excellent performance threshold for the thigh is 4~6kN (thigh bone axial force). Otherwise, the occupant injury mode is determined to be an unstable mode.

[0042] The unstable modes include type B rotational runaway mode, type C translational projectile mode, and type D mixed instability; among them... If the trajectory of the extrapolated centroid index exhibits a rotational characteristic around the vertical axis, and the maximum Lyapunov exponent is greater than or equal to the first stability threshold and less than the second stability threshold, it is determined to be a Class B rotational runaway mode; the second stability threshold is greater than the first stability threshold; specifically, exhibiting a rotational characteristic around the vertical axis means that the rotational angular velocity is greater than the cervical whiplash injury risk threshold. In this embodiment, the cervical whiplash injury risk threshold is 5 rad / s, the first stability threshold is 0.3, and the second stability threshold is 0.6. If the trajectory of the extrapolated center of mass index exceeds the boundary of the virtual support base area within a preset time window, and the maximum Lyapunov exponent is greater than or equal to the second stability threshold, and the longitudinal or lateral velocity of the occupant's center of mass is greater than the secondary collision risk velocity threshold, then it is determined to be a Class C translational projectile mode. The preset time window is 20ms to 50ms. In this embodiment, the preset time window is 30ms. Preferably, in other embodiments, when determining a Class C translational projectile mode, the relationship between the longitudinal or lateral velocity of the occupant's center of gravity and the secondary collision risk velocity threshold can be further used for auxiliary determination. If the longitudinal or lateral velocity of the occupant's center of gravity is greater than the secondary collision risk velocity threshold, it is determined to be a Class C translational projectile mode. Specifically, the secondary collision risk velocity threshold can be obtained by referring to damage thresholds and industry-recognized standard ranges such as those in ISO 61508 and GB / T 20913, and then calibrated based on the aforementioned 100 sets of simulation data; the secondary collision risk velocity threshold can be 2 m / s. Otherwise, it belongs to type D, a mixed unstable type.

[0043] In Class A stable and controllable mode, the human body is effectively controlled by the restraint system, with no obvious tendency to lose control and minor local damage. The local damage level is compared with the excellent performance thresholds recognized by regulations such as Euro NCAP and IIHS. Only when all key local indicators are below these stringent thresholds is the condition for minor local damage met. Typical scenarios include timely seatbelt pretensioning and normal airbag deployment in a frontal collision, with minimal forward thrust and effective pull-back of the human body. In Class B rotational runaway mode, the rotational angular velocity ω_z > 5 rad / s, and 0.3 ≤ λmax < 0.6 (within the critical stability range). In this mode, the human body undergoes lateral or longitudinal rotation, with the cervical and thoracic vertebrae bearing significant torsional loads, easily leading to whiplash injuries or rib fractures. Typical scenarios include delayed side airbag deployment in a side collision, with the human body tilting and rotating significantly towards the collision side.

[0044] In the Class C translational projectile mode, the extrapolated center of mass XcoM exceeds any boundary of the virtual support base area within 30ms after the collision (critical constraint response time); and λmax≥0.6 (in the unstable range); at the same time, the longitudinal (x-axis) or lateral (y-axis) translational velocity of the occupant's center of mass is >2m / s (secondary collision risk speed threshold); at this time, the human body is out of the control of the restraint system and undergoes a large translation, which is very likely to cause a secondary collision with the steering wheel, dashboard, door and other body structures, resulting in severe damage; the typical scenario is that the seat belt is not fastened or slips off in a rear-end collision, and the human body is thrown forward after tilting backward with the seat back.

[0045] In this embodiment, the extrapolated centroid index is analyzed as follows: all three axial components are within the predefined virtual support base area boundary range throughout the entire collision process, with no out-of-bounds situations. The overall motion trend of the occupant is stable. The comparison of local damage indices with the preset excellent performance thresholds shows that all local damage indices of the head, neck, chest, and thighs are lower than the preset excellent performance thresholds, meeting the excellent performance requirements for occupant protection. The comparison of the maximum Lyapunov index with the threshold shows that the maximum Lyapunov index is less than the preset first stability threshold of 0.3. The occupant motion has no trajectory divergence trend, and the local dynamic stability is good. Therefore, it is determined to be a Class A stable and controllable mode.

[0046] Example 2 A virtual vehicle evaluation system based on centroid and Lyapunov index, characterized in that, for running the above method, it includes: The data input module is used to acquire the center of mass motion data of the occupants and the kinematic response data of multiple key parts of the occupants during the vehicle collision. The calculation and analysis module is used to calculate the extrapolated center of mass index and the maximum Lyapunov index based on the center of mass motion data, and to calculate multiple local damage indices based on the kinematic response data of the key parts. The damage mode classification module is used to classify the occupant's damage mode based on the extrapolated centroid index, the maximum Lyapunov index, and multiple local damage indices, according to preset damage mode classification rules.

[0047] The above are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A virtual vehicle evaluation method based on centroid and Lyapunov index, characterized in that, Includes the following steps: S1: Acquire the center of mass motion data of the occupant and the kinematic response data of multiple key parts of the occupant during the vehicle collision. The center of mass motion data includes the center of mass position and center of mass velocity of the occupant. S2: Based on the center of mass motion data, calculate the extrapolated center of mass index for assessing the global dynamic stability of the occupant and the maximum Lyapunov index for assessing the local dynamic stability of the occupant; calculate multiple local injury indices of the occupant based on the kinematic response data of multiple key parts of the occupant. S3: Based on the extrapolated centroid index, the maximum Lyapunov index, and multiple local damage indices, the damage patterns of the occupants are classified according to the preset damage pattern classification rules.

2. The vehicle virtual evaluation method based on centroid and Lyapunov index as described in claim 1, characterized in that: In step S2, the extrapolated centroid index is calculated using the following formula: in, The location of the center of mass. Let be the velocity of the center of mass, h be the height of the center of mass, and g be the acceleration due to gravity.

3. The vehicle virtual evaluation method based on centroid and Lyapunov index according to claim 2, characterized in that: In step S2, the calculation steps for the maximum Lyapunov exponent are as follows: 1) Preprocess and fuse the center-of-mass motion data to construct a fused time series containing position and velocity information; 2) Based on the fused time series, the system dynamics phase space is constructed using the phase space reconstruction method; 3) In phase space, calculate the average exponential divergence rate of neighboring trajectory points over time, and determine the fitting slope of the average exponential divergence rate as the maximum Lyapunov exponent.

4. The vehicle virtual evaluation method based on centroid and Lyapunov index as described in claim 3, characterized in that: When constructing the system dynamics phase space, the CC method is used to determine the delay time, the spurious nearest neighbor method is used to determine the embedding dimension, and the fused time series is reconstructed into an m-dimensional phase space trajectory based on the delay time and the embedding dimension.

5. The vehicle virtual evaluation method based on centroid and Lyapunov index according to claim 3, characterized in that: Determining the maximum Lyapunov exponent specifically includes: In the reconstructed phase space, find the nearest neighbor point that satisfies the distance threshold for each trajectory point to form multiple neighbor pairs; Track the separation distance of each neighboring point pair over the evolution time step, and calculate the average log distance of all point pairs at each evolution step; A linear fit was performed on the curve of the mean logarithmic distance as a function of the evolution step size, and the slope of the fitted line was used as the maximum Lyapunov exponent.

6. The vehicle virtual evaluation method based on centroid and Lyapunov index according to claim 2, characterized in that: In step S1, the key areas include the head, neck, chest, and thigh; in step S2, the local injury indicators include head injury criteria, neck injury indicators, chest compression, and femoral axial force.

7. The vehicle virtual evaluation method based on centroid and Lyapunov index according to claim 1, characterized in that: In step S3, the damage mode classification rule is as follows: The occupant injury mode is determined to be a Class A stable and controllable mode when all three of the following conditions are met simultaneously. a) The trajectory of the extrapolated centroid index lies within the boundary of a predefined virtual support base region throughout the entire collision process; b) The maximum Lyapunov exponent is less than the first stability threshold; c) All local damage indicators are below the preset excellent performance threshold; Otherwise, the occupant injury mode is determined to be an unstable mode.

8. The vehicle virtual evaluation method based on centroid and Lyapunov index according to claim 7, characterized in that: The unstable modes include type B rotational runaway mode, type C translational projectile mode, and type D mixed instability; among them... If the trajectory of the extrapolated centroid index exhibits a rotational characteristic around the vertical axis, and the maximum Lyapunov exponent is greater than or equal to the first stability threshold but less than the second stability threshold, it is determined to be a type B rotational runaway mode. If the trajectory of the extrapolated centroid index exceeds the boundary of the virtual support base region within a preset time window, and the maximum Lyapunov exponent is greater than the second stability threshold, it is determined to be a Class C translational projectile mode. Otherwise, it belongs to type D, a mixed unstable type.

9. The vehicle virtual evaluation method based on centroid and Lyapunov index according to claim 8, characterized in that: The virtual support base area is defined by the spatial range formed by the vehicle seats, seat belt constraint boundaries, and in-vehicle occupant restraint devices.

10. A virtual vehicle evaluation system based on centroid and Lyapunov index, characterized in that, To run the above methods, the following are included: The data input module is used to acquire the center of mass motion data of the occupants and the kinematic response data of multiple key parts of the occupants during the vehicle collision. The calculation and analysis module is used to calculate extrapolated centroid indices and maximum Lyapunov index based on centroid motion data, and to calculate multiple local damage indices based on kinematic response data of key parts. The damage mode classification module is used to classify the occupant's damage mode based on the extrapolated centroid index, the maximum Lyapunov index, and multiple local damage indices, according to preset damage mode classification rules.

Images