CN1889083A - Car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model - Google Patents

Abstract

An optimizing method for the colliding accident of car-human based on the three-dimensional multi-steel model. The process: (1) make the three-dimensional multi-steel mathematical model of the car, the human and the environment; (2) determine the centroid position coordinate of the car stopping and the coordinate of the human buttocks; (3) determine the optimizing design variable, the limiting condition and the optimizing object function; (4) resolve the giving object function by the optimizing arithmetic; (5) determine the astringency of the function after computing. The invention has decreases the reappearance time, the error and the indeterminacy by man-made interference.

Description

Car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model

Technical field

The present invention relates to a kind of collision accident recurring optimizing method that is used for field of traffic, specifically is a kind of car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model.

Background technology

Along with increasing sharply of China's automobile pollution, add relatively lagging behind of road traffic infrastructure construction, it is frequent to cause the Traffic Collision accident to take place, and wherein the collision accident proportion of automobile and pedestrian generation is higher, accounts for more than 1/4th of sum.Because complicacy that the actual collision accident takes place and the on-the-spot diversity of leaving over information have strengthened the difficulty of accident being reproduced analysis, are unfavorable for providing foundation fast and accurately for correctly evaluation accident of vehicle supervision department.

The key of accident reconstruction is to calculate definite Automobile Collision Velocity according to the information of leaving over of the scene of the accident, accurately reproduces whole collision process with this.For automobile and pedestrian impact accident, common reproduction research method has: momentum/Energy Analysis for High, distortion/Energy Analysis for High, projected object are thrown pitch analysis method etc.Momentum/Energy Analysis for High mainly relies on the braking automobile trace, and automobile and pedestrian are reduced to two-dimentional lumped mass model, and accuracy is not high, simultaneously along with the widespread usage of ABS technology on automobile, is difficult to obtain the brake trace of complete display; Distortion/Energy Analysis for High is considered the body of a motor car collision deformation, but the non-linear and elastoplasticity of material makes that the linear approximate relationship between body distortion and the impact velocity is only effective within the specific limits, the resource cost that utilizes finite element simulation to calculate simultaneously is huge, and efficient is low; Projected object is thrown pitch analysis method and is stopped or drop point when uncertain running into when ejecta, can produce than the computation error; As seen, the limitation of above-mentioned analytical approach is unfavorable for the pedestrian-vehicle collision accident of complexity is reproduced analysis.

Along with digitizing dummy's research and development in recent years, Yang etc. are at " International Journal ofCrash prevention and Injury Control " 2000, Vol.2 (2): " the AHuman-Body 3D Mathematical Model for Simulation of Car-Pedestrian Impacts " that delivers on the 131-149, (" a kind of three-dimensional numerical value dummy model that is used for automobile to the simulation study of pedestrian impact accident ", " international crash protection and damage Control magazine ", 2000, Vol.2 (2): 131-149) early having set up three-dimensional pedestrian's numerical value dummy model carries out the computer simulation of pedestrian-vehicle collision accident in the literary composition, have fidelity and accuracy preferably, but its emulation is reproduced process need and is artificially constantly calculated debugging according to the single calculation result, non-Automatic Program optimization is finished, rely on individual's experience to a great extent, cause error easily, increase the time that accident reconstruction is analyzed, reduced efficient.

Therefore, how more accurately, expeditiously the pedestrian-vehicle collision accident of numerous complicated is reproduced analysis, be a problem that presses for solution that faces.

Summary of the invention

At the problem that present accident reconstruction exists, the present invention proposes a kind of car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model.The present invention is optimized collision process, has shortened the recovery time, has reduced the sum of errors uncertainty that human intervention brings, can be more accurately, expeditiously car-people's collision accident is reproduced analysis.

The present invention is achieved through the following technical solutions, and concrete steps comprise:

(1) sets up three-dimensional many rigid bodies mathematical model of accident generation environment and accident car, pedestrian;

According to the actual accidents situation, set up the three-dimensional many Rigid Body Collision mathematical model that comprises road surface, guardrail, car, pedestrian, its equation of motion is:

$\mathrm{\Σ}[{\mathrm{\δ\γ}}_i(m_i{\stackrel{\·\·}{r}}_i-F_i)+{\mathrm{\δ\π}}_i(J_i{\stackrel{\·}{w}}_i+w_i{\×J}_i{w}_i-T_i)]=0---(i=1,...,n)$

M in the following formula _iBe quality,

Be acceleration, J _iBe moment of inertia, w _iBe angular velocity, F _iBe power, T _iBe moment, δ γ _iBe offset variable, δ π _iBe the initial orientation variable.Can determine the collision contact stiffness characteristic of ground, guardrail and car by impact test, the mechanical characteristic of pedestrian dummy is set according to the body biomechanics test figure.

(2) determine that accident the back car takes place stops centroid position coordinate, pedestrian's buttocks drop point site coordinate;

According to scene of the accident environment, a self-defined fixing space generalized coordinate is XYZ, can determine that then accident the back car takes place stops centroid position coordinate C ₁(x _C1, Y _C1, z _C1), pedestrian's buttocks drop point site coordinate H ₁(x _H1, y _H1, z _H1).

(3) determine optimal design variable, constraint condition and optimization aim function;

Car impact velocity and collision initial time position are known variables in the accident, are the impact velocity V of car to the pedestrian so set the optimal design variable, the collision initial time car centroid position coordinate C in coordinate system XYZ ₂(x _C2, y _C2, z _C2), the pedestrian buttocks position coordinates H that stands ₂(x _H2, y _H2, z _H2).

Leave over information according to the actual accidents scene, according to a preliminary estimate the interval of car impact velocity V, coordinate points C simultaneously ₂And H ₂On the travel direction of car with respect to coordinate points C ₁And H ₁There is certain distance relation, determines the constraint condition of design variable by this relation.

With coordinate points C ₁, H ₁Be the optimization aim point, determine the optimization aim function:

$Y=F\left[x\right]=\mathrm{Min}\underset{i=0}{\overset{m-1}{\mathrm{\Σ}}}{\mathrm{\β}}_{i}f\left(x_i\right)$

β in the formula _iBe weighting coefficient, generally can be taken as 1 entirely.

(4) calling optimized Algorithm is optimized given objective function and finds the solution;

Call optimized Algorithm, in design variable interval scope, search for, after p suboptimization circulation, calculate impact velocity V ^p, car stops centroid position coordinate C ₁ ^p(x _C1 ^p, y _C1 ^p, z _C1 ^p), pedestrian's buttocks drop point site coordinate H ₁ ^p(x _H1 ^p, y _H1 ^p, z _H1 ^p), make objective function F ^p[x] minimizes, that is:

${\mathrm{MinF}}^p\left[x\right]=\mathrm{Min}({[{(p_{c1}^p-x_{c1})}^2+{(y_{c1}^p-y_{c1})}^2+{(z_{c1}^p-z_{c1})}^2]}^{1/2}+{[(x_{h1}^p-x_{h1})}^2+{(y_{h1}^p-y_{h1})}^2+{(z_{h1}^p-z_{h1})}^2{]}^{1/2})$

(5) after repeatedly optimizing cycle calculations, judge the objective function convergence, whether try to achieve optimum solution.

Through p suboptimization cycle calculations, satisfying under the situation of constraint condition, calculation procedure compares analysis to the result of calculation that every adjacent two optimization circulations obtain, according to the objective function error E of setting ^pJudge whether convergence.Objective function Error Calculation formula is:

$E^p=\left|\frac{F^P\left[x\right]-F^{P-1}\left[x\right]}{F^{P-1}\left[x\right]}\right|$

E ^pInterval is [0,1], increases E along with optimizing number of times p ^pMore little, show the objective function error sequence trend that tapers off, twice adjacent optimization round-robin result of calculation is approaching more, optimizes the result and is tending towards convergence more.Work as E ^pValue is enough little in the value range of setting, and can judge the objective function convergence, tries to achieve optimum solution.

The present invention carries out accident reconstruction by rigid multibody dynamics method and optimization method, have following advantage: 1, this method is set up accident environment, car, pedestrian's three-dimensional impact model, collision process is more directly perceived, has avoided car and pedestrian are reduced to defectives such as the error of calculation, visual degree that two dimensional surface lumped mass model brings be low; 2, utilize the rigid multibody dynamics method to carry out collision accident reappearance, efficient is higher, has avoided Finite Element Method to expend shortcomings such as computational resource is big, reproduction length computing time; 3, determine design variable and optimization aim after, calculation procedure is finished the optimization of collision process automatically and is reproduced, human intervention is little, reproduces objectivity degree height as a result; 4, but this method analog computation goes out each position of pedestrian and is hurt in collision process, is convenient to be analyzed with the accident actual conditions, can provide reference and theoretical the support for legal medical expert identifies simultaneously.

Description of drawings

Fig. 1 is the three-dimensional multi-rigid model of accident environment, car, pedestrian in the embodiment of the invention

Fig. 2 is car in the embodiment of the invention-people's collision accident scene photo

Fig. 3 is an embodiment of the invention design sketch

Embodiment

The present invention is according to the information of leaving over of car and the pedestrian impact accident characteristic and the scene of the accident, and the selection car stops centroid position and pedestrian's buttocks drop point site is an optimization aim, with the objective function error value E ^pBe evaluation index, it has reflected whether the result of calculation that p suboptimization circulation obtains is tending towards convergence.

According to fixing three dimensional space coordinate of scene of the accident definable is XYZ, determines that then accident the back car takes place stops centroid position coordinate C ₁(x _C1, y _C1, z _C1), pedestrian's buttocks drop point site coordinate H ₁(x _H1, y _H1, z _H1).

With the impact velocity V of car to the pedestrian, collision initial time car centroid position coordinate C ₂(x _C2, y _C2, z _C2), the pedestrian buttocks position coordinates H that stands ₂(x _H2, y _H2, z _H2) be design variable.Calculate by p optimizing of computing machine, can obtain impact velocity V ^p, car stops centroid position coordinate C ₁ ^p(x _C1 ^p, y _C1 ^p, z _C1 ^p), pedestrian's buttocks drop point site coordinate H ₁ ^p(x _H1 ^p, y _H1 ^p, z _H1 ^p), make objective function F [x] minimum, that is:

${\mathrm{MinF}}^p\left[x\right]=\mathrm{Min}({[{(x_{c1}^p-x_{c1})}^2+{(y_{c1}^p-y_{c1})}^2+{(z_{c1}^p-z_{c1})}^2]}^{1/2}+{[(x_{h1}^p-x_{h1})}^2+{(y_{h1}^p-y_{h1})}^2+{(z_{h1}^p-z_{h1})}^2{]}^{1/2})$

F[x] more little, the approaching more true accident conditions of result of calculation are described, C ₁ ^p, H ₁ ^pRespectively with C ₁, H ₁Approach to overlap.

Because there is certain limitation in the mathematical optimization method, optimization result who obtains and truth can not be in full accord, E ^pReaction is optimized the result and whether is tending towards convergence, and its value reduces gradually, illustrates that optimizing the result is convergent tendency, can finally try to achieve F[x] minimum value.

Provide specific embodiment below in conjunction with content of the present invention:

With a real car and pedestrian impact accident is example, and the collision accident scene as shown in Figure 2.With the car centroid position is that true origin is set up fixing space coordinates XYZ (unit: m), then, can determine that accident the back car takes place stops centroid position coordinate (0,0,0), pedestrian's buttocks drop point site coordinate (10.25 ,-4.66 ,-0.51).

According to braking trace length, the car travel speed is between 30km/h～80km/h according to a preliminary estimate, and collision initial time car and pedestrian's position can be moved between 0～40m distance to the X negative direction along braking trace in coordinate system XYZ.

Utilize car and the three-dimensional many Rigid Body Collision model of pedestrian set up among Fig. 1, is the optimal design variable with car to pedestrian impact speed and collision moment position coordinates, car stops centroid position and pedestrian's buttocks drop point site is an optimization aim, calls optimized Algorithm and carries out optimizing calculating.

By repeatedly optimizing cycle calculations, trying to achieve car and the pedestrian impact moment speed of a motor vehicle at last is 41km/h, and car stops center-of-mass coordinate (0.42 ,-0.13,0), pedestrian's buttocks landing point coordinates (10.01 ,-4.97 ,-0.56), and the fine reproduction accident sight of energy, as shown in Figure 3.This moment objective function F [x]=0.835, the objective function error E ^p=0.0032, illustrate that the optimization result is approaching with true accident, the utilization optimization method has carried out reproducing well analysis to collision accident.

Claims (6)

1, a kind of car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model is characterized in that:

(1) sets up three-dimensional many rigid bodies mathematical model of accident generation environment and accident car, pedestrian;

(2) determine that accident the back car takes place stops centroid position coordinate, pedestrian's buttocks drop point site coordinate;

(3) determine optimal design variable, constraint condition and optimization aim function;

(4) calling optimized Algorithm is optimized given objective function and finds the solution;

(5) after repeatedly optimizing cycle calculations, judge the objective function convergence, whether try to achieve optimum solution.

2, the car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model according to claim 1, it is characterized in that, described step (1), specifically refer to: for real car-people's collision accident, according to the accident actual conditions, set up the three-dimensional multi-rigid model of the accident generation environment that comprises road surface, guardrail; Set up three-dimensional multi-rigid model according to automobile geometric shape feature; Consideration pedestrian height, age, body weight are set up three-dimensional multi-rigid model.

3, the car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model according to claim 1, it is characterized in that, described step (2), specifically refer to: according to scene of the accident environment, self-defined space generalized coordinate is XYZ, determines that then accident the back car takes place stops centroid position coordinate C ₁(x _C1, y _C1, z _C1), pedestrian's buttocks drop point site coordinate H ₁(x _H1, y _H1, z _H1).

4, the car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model according to claim 1, it is characterized in that, described step (3) specifically refers to: setting the optimal design variable is the impact velocity V of car to the pedestrian, collision initial time car centroid position coordinate C ₂(x _C2, y _C2, z _C2), the pedestrian buttocks position coordinates H that stands ₂(x _H2, y _H2, z _H2), and distance relation that car, people collide front and back interval with the valuation of car speed determined the optimization aim function as constraint condition:

$Y=F\left[x\right]=\mathrm{Min}\underset{i=0}{\overset{m-1}{\mathrm{\Σ}}}{\mathrm{\β}}_{i}f\left(x_i\right)$

β in the formula _iBe weighting coefficient.

5, the car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model according to claim 1, it is characterized in that, described step (4), specifically refer to: call optimized Algorithm, in design variable interval scope, search for, after p suboptimization circulation, calculate impact velocity V ^p, car stops centroid position coordinate C ₁ ^p(x _C1 ^p, y _C1 ^p, z _C1 ^p), pedestrian's buttocks drop point site coordinate H ₁ ^p(x _H1 ^p, y _H1 ^p, z _H1 ^p), make objective function F ^p[x] minimizes, that is:

$\mathrm{Min}{F}^p\left[x\right]=\mathrm{Min}({[{(x_{c1}^p-x_{c1})}^2+{(y_{c1}^p-y_{c1})}^2+{(z_{c1}^p-z_{c1})}^2]}^{1/2}+{[{(x_{h1}^p-x_{h1})}^2+{(y_{h1}^p-y_{h1})}^2+{(z_{h1}^p-z_{h1})}^2]}^{1/2})$

。

6, the car-human collision accident recurring optimizing method based on three-dimensional multi-rigid model according to claim 1, it is characterized in that, described step (5), specifically refer to: through p suboptimization cycle calculations, satisfying under the situation of constraint condition, calculation procedure compares analysis to the result of calculation that every adjacent two optimization circulations obtain, according to the objective function error E of setting ^pJudge whether convergence.Objective function Error Calculation formula is:

$E^p=\left|\frac{F^P\left[x\right]-F^{P-1}\left[x\right]}{F^{P-1}\left[x\right]}\right|$

E ^pInterval is [0,1], increases E along with optimizing number of times p ^pMore little, show the objective function error sequence trend that tapers off, twice adjacent optimization round-robin result of calculation is approaching more, optimizes the result and is tending towards convergence more.Work as E ^pValue is enough little in the value range of setting, and can judge the objective function convergence, tries to achieve optimum solution.

Images