CN101714175B - Optimization theory-based automobile flanging die edge retraction control method - Google Patents

Abstract

The invention discloses an optimization theory-based automobile flanging die edge retraction control method, which comprises the following steps: 1) selecting Rd, H and G as design parameters of edge retraction P after edge hemming of plates; 2) performing parameter conversion on the design parameters so as to eliminate a magnitude relation therebetween; 3) sampling by a Latin hypercube experiment design method in a space consisting of the design parameters of Rd, H and G to obtain a series of sample points capable of reflecting characteristics of the whole sampling space; 4) calculating the sample points with finite element software to obtain the edge retraction P of the sample points; 5) establishing a radial basis function-based approximate model according to the edge retraction P; and 6) performing deterministic optimization with the approximate model to obtain the optimal combination of the design parameters, which makes the edge retraction P minimal. The method makes the design of a flanging die more easy and computational efficiency greatly improved so as to greatly reduce design cost of the flanging die; and the method is easy to grasp.

Description

A kind of automobile bound edge mould border retraction control method based on optimum theory

Technical field

The present invention relates to a kind of bound edge mould border indentation control technology, particularly a kind of automobile bound edge mould border retraction control method based on optimum theory.

Background technology

The left and right sides car door of automobile, front and back door, hood and luggage-boot lid, promptly four-door two-lid is the important component part of body of a motor car assembly, also is the parts that need bound edge on the automobile.When bound edge, treat that bound edge inwardly turns down 90 ° and encases interior panel edges.Theoretical analysis and test show, is difficult to accomplish with treating that bound edge encases interior panel edges from 90 ° of one step turnover to 0 °.Therefore, its wrapping process (referring to figs. 1 through Fig. 3) comprising: locate, compress (shown in Figure 1) earlier; To treat bound edge from 90 ° of turnovers to 45 ° (shown in Figure 2), and then with it from 45 ° of turnovers to 0 ° and compacting (shown in Figure 3).What generally adopt at present is mould pressing bound edge, accomplishes 45 ° of bound edges and 90 ° of bound edges simultaneously through adopting a mold.The plate bound edge is last procedure that the four-door two-lid parts of automobile are produced, so the quality of bound edge quality can have very big influence to the quality of product.And whether the border indentation P (referring to Fig. 4, Fig. 5) behind the plate bound edge evenly has very big influence to assembling behind the auto sheet feeder bound edge and the slit between the plate, is the problem that enterprise extremely is concerned about so how to make the border indentation for minimum.But because the complicacy of auto sheet feeder bound edge mould; Influence the uncertainty that parameter of border indentation behind the plate bound edge; And the sheet metal deformation zone is little; When the border indentation that utilizes business software to calculate the auto sheet feeder bound edge computing time very long, and can not be optimized design to the border indentation behind the plate bound edge, the control to the border indentation in actual production all is the experience that relies on deviser and model commissioning staff; Just caused bound edge mould border indentation control difficulty big, the cost of its design and debugging is high.Do not have as yet at present that a kind of efficient is high, low, the wield border indentation of cost control technology comes designer to the bound edge mould to instruct.

Summary of the invention

Problem to be solved by this invention provides a kind of efficient height, low, the wield automobile bound edge mould border retraction control method based on optimum theory of cost.

The technical scheme that the present invention adopted is: the automobile bound edge mould border retraction control method based on optimum theory may further comprise the steps:

1. select flange knuckle radius R _d, the clearance G between the stile limit height H, inner and outer plates as the plate bound edge after the design parameter of border indentation P;

2. said design parameter is carried out parameter and transform, to eliminate the magnitude relationship between them;

With the Latin hypercube experimental design method at above-mentioned design parameter R _d, H, G sample in the space formed, obtains reflecting a series of sample points of whole sample space characteristic;

4. with finite element software said sample point is calculated, obtain at the border of sample point indentation P;

5. set up approximate model according to the border indentation P that obtains based on RBF;

6. carry out determinacy optimization with approximate model, obtain making border indentation P to reach the optimum combination of the said design parameter of minimum.

Characteristics of the present invention are through selecting to influence the susceptibility design parameter of indentation P in border behind the plate bound edge: flange knuckle radius R _d, the clearance G between the stile limit height H, inner and outer plates; Set up the RBF approximate model of plate bound edge; And replace true model to be optimized with approximate model; Acquisition makes the border indentation reach the optimum combination of the design parameter of minimum, makes Optimization result in actual production, to use, in order to instruct the designer of bound edge mould.The inventive method makes the design of bound edge mould be more prone to, and counting yield improves greatly, thereby can reduce the design cost of bound edge mould greatly, and easy master.

Description of drawings

Fig. 1 to Fig. 3 is a plate wrapping process synoptic diagram;

Fig. 4, Fig. 5 are the border indentation P synoptic diagram of plate behind bound edge;

Fig. 6 is the flange knuckle radius R of plate when bound edge _d, the clearance G synoptic diagram between the stile limit height H, inner and outer plates;

Fig. 7 is the synoptic diagram that concerns that bound edge plate design parameter and process transform each parameter of back.

Embodiment

The present invention includes following steps:

1, see Fig. 6, the inner panel of plate 1 and outside plate 2 are when bound edge, and the design parameter of border indentation (P) carries out single sensitive analysis of factors behind the plate bound edge to influencing, and finds three design parameters that influence the border indentation of bound edge plate: flange knuckle radius R _d, the clearance G between the stile limit height H, inner and outer plates, wherein 0.5mm≤R _d≤4.0mm, 0.5mm≤G≤4.0mm, 3.0mm≤H≤16.0mm;

2, because of existence condition: R in these design parameters _d≤G≤H so need the design parameter of bound edge plate be transformed, should not have magnitude relationship between the sample parameter of in Latin hypercube experimental design method method is sampled, being adopted, thus need the design parameter of bound edge plate be transformed,

H＝G+A (1)

G＝R _d+B (2)

Can the design parameter of plate bound edge be converted into A, R by formula (1) and formula (2) _dAnd B, wherein A is illustrated in outside plate and the equitant part of inner panel behind the plate bound edge, and B just representes the distance (as shown in Figure 7) in the stile inner arc root and the inner panel outside, has not just had magnitude relationship between these three parameters.Certainly also can adopt other parameter method for transformation.

3. with Latin hypercube experimental design method (Morris M.D.; Mitchell T.J.; Exploratory designsfor computational experiments.Journal of Statistical Planning and Inference; 1995,43 (3): 381-402.) at above-mentioned three design parameter A, R _dSample in the space of being formed with B and obtain sample point (as shown in table 1);

4. with finite element software Ls-dyna sample point is calculated, obtain the border indentation P value of sample point, see table 1;

5. according to the design parameter after transforming and finite element software Ls-dyna result of calculation at sampled point; Foundation is based on the approximate model (Mu Xuefeng of RBF; Yao Weixing, Yu Xiongqing etc. in the multidisciplinary design optimization+and the research of agent model commonly used. Computational Mechanics journal, 2005; 22 (5): 608-612.), the citation form of the radially basic mode type of the response f of tested point x place (x) is following:

$f_p\left(x\right)=\underset{i}{\overset{n}{\mathrm{\Σ}}}{w}_i\mathrm{\Φ}\left(r^i\right)---\left(3\right)$

Wherein, n is a number of samples, w _i(i=1,2 ...) be weight coefficient, r ⁱ=|| x-x _i|| be the Euclidean distance between estimation point and the sample point, Φ (r) is radial function (apart from the monotonic quantity of r), and present embodiment selects for use the Gauss function as RBF:

Φ(r)＝exp(-r ²/c ²)(4)

C is given greater than zero constant.According to the interpolation principle, the corresponding relation when knowing n sample point and each sample point corresponding response functional value can obtain following system of equations, is write as matrix form:

f＝Φw (5)

Can obtain n corresponding weight coefficient:

w＝Φ ^-1f (6)

6. use by bound edge plate design parameter and the finite element software Ls-dyna border indentation after approximate model that the result of calculation (as shown in table 1) of sampled point is set up is to the plate bound edge and carry out determinacy optimization (Xu; Y.G., Liu G, R.; Wu; Z.P., 2001.A novel hybrid genetic algorithm using localoptimizer based on heuristic pattern more.Applied ArtificalIntelligence 15,601-631.)

$\underset{X}{\min }f\left(X\right)---\left(7\right)$

X∈[X ^L，X ^R]

X representes n dimension design vector, and span is Ω ⁿF representes objective function, is the continuous nonlinear function about X, and subscript L and R have represented the lower bound and the upper bound of interval number respectively.

Calculating the optimum combination that makes the border indentation reach three responsive design parameters of minimum by above step is A=2.5127mm, R _d=0.5mm, B=0mm brings these three parameters in formula (1) and (2) into and to obtain H, G and R _dDesign load as shown in table 2.Set up the emulation border indentation that finite element model calculates finite element according to these three design loads; And with optimize the border indentation and compare; Its result is as shown in table 2, and is not little through contrast display optimization border indentation and emulation border indentation value difference, proves that it is believable optimizing the border indentation.

Table 1 sampled point parameter and in the result of calculation of sampled point border indentation

A(mm)	R d(mm)	B(mm)	P(mm)
				5.1910	0.8500	0.4710	0.5591
3.3385	1.1510	0.7290	0.7068
				5.5615	3.9510	0.8130	1.6740
5.7500	0.9515	2.6130	0.6898
				4.4500	2.6490	1.4130	1.1395
2.9615	3.3490	2.0130	1.4910
				7.0500	1.8510	1.5870	0.9117
6.3090	2.9500	0.0420	1.1507
				3.5205	3.6500	1.0710	1.6190
4.6385	0.5490	1.5000	0.5230
				8.1615	2.3515	0.3870	0.9495
3.8910	3.1495	0.3000	1.3550
				6.1205	2.5510	2.2710	1.1043
7.7910	3.7515	1.8420	1.6290
				8.9090	1.6515	2.1000	0.8533
7.4205	1.7495	2.8710	0.9082
				8.3500	2.8485	2.4420	1.3621
8.5385	1.3505	0.9870	0.7429
				5.3795	3.8495	1.9290	1.6460
4.0795	1.9490	0.1290	0.9214
				4.8205	2.1485	2.9580	1.0120
6.4910	3.0515	1.2420	1.2581
				6.6795	3.4505	2.7870	1.4810
3.1500	2.2500	2.3580	1.0423
				7.6090	0.7485	2.1870	0.5934
4.2615	3.2510	2.7000	1.3801
				8.7205	2.7505	1.3290	1.1360
7.2385	1.2490	0.2130	0.6502
				2.7795	1.4485	1.6710	0.8408
2.5910	2.4495	0.9000	1.1447
				3.7090	1.0495	2.5290	0.7458
6.8615	0.6505	1.1580	0.5271
				5.0090	1.5500	1.7580	0.8756
5.9385	2.0505	0.6420	0.9560
				7.9795	3.5485	0.5580	1.4010

Table 2 determinacy is optimized border indentation and emulation border indentation contrast

Claims (1)

1. automobile bound edge mould border retraction control method based on optimum theory is characterized in that carrying out according to the following steps:

1) selects flange knuckle radius R _d, the clearance G between the stile limit height H, inner and outer plates as the plate bound edge after the design parameter of border indentation P;

2) said design parameter is carried out parameter by formula (1) and formula (2) and transform, to eliminate the magnitude relationship between them;

H＝G+A (1)

G＝R _d+B (2)

In the formula: A is illustrated in the length of outside plate and the equitant part of inner panel behind the plate bound edge, and B representes the distance in the stile inner arc root and the inner panel outside;

3) sample and obtain sample point in the space of with the Latin hypercube experimental design method this resulting design parameter in parameter conversion back being formed;

4) with finite element software Ls-dyna said sample point is calculated, obtain at the border of sample point indentation P;

5) according to design parameter after transforming and the result of calculation of finite element software Ls-dyna at sample point, foundation is based on the approximate model of RBF, the response f of tested point x place _pThe citation form of radially basic mode type (x) is formula (3):

$f_p\left(x\right)=\underset{i}{\overset{n}{\mathrm{\Σ}}}{w}_i\mathrm{\Φ}\left(r^i\right)---\left(3\right)$

In the formula, n is a number of samples, w _i(i=1,2 ..., n) be weight coefficient, r ⁱ=|| x-x _i|| be the Euclidean distance between tested point and the sample point, Φ (r ⁱ) be that radial function is promptly apart from r ⁱMonotonic quantity, select for use the Gauss function as RBF, formula (4) is then arranged:

Φ(r)＝exp(-r ²/c ²) (4)

In the formula, c is given greater than zero constant, and according to the interpolation principle, the corresponding relation when knowing n sample point and each sample point corresponding response functional value can obtain following system of equations, is write as matrix form (5):

f＝Φw (5)

F representative function f wherein _p(x), Φ represents radial function Φ (r ⁱ), and w representation coefficient can use formula (6) to obtain n corresponding weight coefficient:

w＝Φ ^-1f (6)

6) carry out determinacy optimization with approximate model, design parameter A, R after the conversion that obtains making border indentation P reach minimum _dWith the optimum combination of B, obtaining design parameter H, G and R in these three parameter substitution design parameter conversion formulas (1) and (2) _dDesign load.

Images