CN103136424A - Universal design structure mining oriented method for quantitatively describing assembly models - Google Patents

Abstract

The invention provides a universal design structure mining oriented method for quantitatively describing assembly models. The universal design structure mining oriented method includes steps of firstly, acquiring node sets V and edge sets E of corresponding GFAGs (generic face adjacent graphs) from the assembly models; secondly, quantizing geometrical element information; thirdly, quantizing models of unit parts; fourthly, quantizing the edge sets E; and fifthly, quantizing the node sets V of the GFAGs to complete a procedure for quantitatively describing the assembly models. The universal design structure mining oriented method has the advantages that the assembly models are converted into the GFAGs, quantized coordinates [x(v), y(v)] of various elements of the node sets V corresponding to the assembly models represent nodes of the GFAGs in a one-to-one correspondence manner, a connecting line between each two nodes represents assembly constraint between two corresponding parts, accordingly, topological information of the assembly models is captured by the GFAGs, the geometrical information of the assembly models are quantitatively described, and the topological information can be compared to the geometrical information without further conversion.

Description

Assembling model quantitative description towards the excavation of universal design structure

Technical field

The invention belongs to field of computer aided design, specifically a kind of assembling model quantitative description of excavating towards the universal design structure.

Background technology

The Product Assembly model is the basis of the technology such as assembling process of products planning, path planning, assembling capacity analysis, usually adopts Graph-theoretical Approach to describe qualitatively.In recent years, in order automatically to identify the universal design structure from product model, realize the customization of automatic mining and the Feature library of design knowledge, a kind of universal design structure technique of excavation occurred in field of computer aided design, and this technology need to be quantitatively described to the topological sum geological information of model usually.Current, obtaining some achievements aspect the excavation of part model universal design structure:

At document (Ma LJ, Huang ZD, Wang YW.Automatic discovery of common design structures in CAD models.Computers﹠amp; Graphics, 2010; 34:545-555.) in a kind of adjacent map (face adjacency graph has been proposed, FAG) technology, be node and limit in FAG with the B-rep Model Mapping of part, and caught quantitatively the form parameter of model with the FAG node, both catch the topology information of model, caught again the geological information of model.But this technology only limits to describe part model, and further expansion is not done in the description of assembling model.

Document (Deshmukh AS, Banerjee AG, Gupta SK, Sriram RD.Content-based assembly search:A step towards assembly reuse.Computer-Aided Design, 2008; 40:244-261.) towards the assembling model retrieval, a kind of match map has been proposed.In this figure, node is corresponding to a part, wherein comprised the information such as generic, how much, type of part.Match map can be caught the topology information of assembling model and certain geological information.But the geological information in figure is a kind of nominal pointer, only is used for showing whether a certain part is standard component or customization part.Therefore, the method lacks the description to the assembling model geological information, more can't obtain a kind of quantitative description that can be used for excavating universal design structure in assembling model.

Document (Chen X, Gao S, Guo S, Bai J.A flexible assembly retrieval approach for model reuse.Computer-Aided Design, 2012; 44:554-574.) in the flexible assembly search method of its proposition, developed a kind of multi-level assembling model descriptor.This descriptor comprehensive description semantic, the geological information of the topology information of assembling model, assembling, and other Useful Informations.Geological information wherein adopts the method for distribution of shapes vector to calculate.This descriptor can be described quantitatively to assembling model, but this quantitative description is a kind of distribution plan, also needs it is further changed when analyzing.

The above analysis, current there is no is directly used in the assembling model describing method that the universal design structure is excavated, existing model description method or be confined to part model, or lack geological information, maybe need do further conversion, therefore be badly in need of a kind of applicable describing method of exploitation.

Summary of the invention

The technical matters that solves

For prior art or be confined to part model or lack geological information describe, maybe need do the further deficiency of conversion, the present invention proposes a kind of assembling model quantitative description of excavating towards the universal design structure, excavate service for the universal design structure, improve reuse efficiency and the quality of assembling model.

Technical scheme

The present invention adopts broad sense face adjacent map (Generic Face Adjacent Graph, GFAG) that assembling model is quantitatively described, and by by the graph-based mining algorithm, just can realize the excavation of universal design structure in assembling model on this basis.

Technical scheme of the present invention is:

Described a kind of assembling model quantitative description of excavating towards the universal design structure is characterized in that: adopt following steps:

Step 1: the set of node V and limit collection E, wherein the set of node V={v that generate broad sense face adjacent map according to the part in assembling model ₁, v ₂... } and in element v _iCorresponding to a part in assembling model, limit collection E={e ₁, e ₂... } and in element e _iCorresponding to the assembly restriction between certain two node;

Step 2: set up the geometric element information quantization model of each geometric surface of part in assembling model, for geometric surface f, its geometric element information is determined by following formula:

$\left\{\begin{array}{c}x\left(f\right)=\underset{i=1}{\overset{n_l}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i[p_x\left(f\right)+p_{e,x}(f,f_i)][p_x\left(f_i\right)+p_{e,x}(f,f_i)]\\ y\left(f\right)=\underset{i=1}{\overset{n_l}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i[p_y\left(f\right)+p_{e,y}(f,f_i)][p_y\left(f_i\right)+p_{e,y}(f,f_i)]\end{array}\right.$

Wherein, f is current geometric surface, f _iBe i the adjacent surface of f, n _lQuantity for the adjacent surface of f; ε _i=(T (f) T (f _i)/θ (f, f _i)), T (f) means the integer of Noodles type: corresponding T (f)=1 when f is the plane, corresponding T (f)=2 when f is the face of cylinder, corresponding T (f)=3 when f is circular conical surface, corresponding T (f)=4 when f is sphere, corresponding T (f)=5 when f is anchor ring, corresponding T (f)=6 when f is free form surface; θ (f, f _i) represent face f and f _iBetween two subangles, p (f) and p _e(f, f _i) be the form parameter of describing geometric surface and geometrical edge, p _x, p _y, p _e,x, p _{E, y}The x component and the y component that represent respectively the respective shapes parameter;

Step 3: the geometric element information quantization model of each geometric surface of each part in the assembling model that obtains according to step 2, set up the quantitative model of single part model:

$\left\{\begin{array}{c}x\left(p\right)=\underset{i=1}{\overset{n_f}{\mathrm{\Σ}}}x\left(f_i\right)/n_f\\ y\left(p\right)=\underset{i=1}{\overset{n_f}{\mathrm{\Σ}}}y\left(f_i\right)/n_f\end{array}\right.$

Wherein p represents part, and x (f) and y (f) are calculated by step 2, n _fSum for geometric surface in part model p;

Step 4: the geometric element information quantization model of each geometric surface of each part in the assembling model that obtains according to step 2, set up the quantitative model of assembly constraint between two parts:

$\left\{\begin{array}{c}x\left(e\right)=x(p_i,p_j)=\underset{k=1}{\overset{n_e}{\mathrm{\Σ}}}\frac{T(f_{i,k},f_{j,k})[x\left(f_{i,k}\right)+x\left(f_{j,k}\right)]}{2n_e}\\ y\left(e\right)=y(p_i,p_j)=\underset{k=1}{\overset{n_e}{\mathrm{\Σ}}}\frac{T(f_{i,k},f_{j,k})[y\left(f_{i,k}\right)+y\left(f_{j,k}\right)]}{2n_e}\end{array}\right.$

F wherein _i,kAnd f _j,kExpression part p _i, p _jBetween the k that cooperatively interacts to geometric surface; T(f _i, f _j) be the fiting constraint type of geometric surface, " being harmonious " retrains corresponding T (f _i, f _j)=1, " contact " retrain corresponding T (f _i, f _j)=2, " skew " retrain corresponding T (f _i, f _j)=3, " angle " retrain corresponding T (f _i, f _j)=4; n _eBe two part p _i, p _jBetween the logarithm of the geometric surface that cooperatively interacts; X (f) and y (f) are calculated by step 2;

Step 5: the quantitative model of assembly constraint between the quantitative model of the single part model that obtains according to step 3 and two parts that step 4 obtains, set up the quantitative model of each element in set of node V:

$\left\{\begin{array}{c}x\left(v\right)=\underset{i=1}{\overset{n_p}{\mathrm{\Σ}}}[x\left(p\right)+x(p,p_i)][x\left(p_i\right)+x(p,p_i)]\\ y\left(v\right)=\underset{i=1}{\overset{n_p}{\mathrm{\Σ}}}[y\left(p\right)+y(p,p_i)][y\left(p_i\right)+y(p,p_i)]\end{array}\right.$

Wherein p is current part model, p _iBe i the part model that coordinates with current part, n _pQuantity for mating parts; X (p) and y (p) are calculated by step 3; X (p, p _i) and y (p, p _i) calculated by step 4;

Step 6: in the set of node V that step 5 is obtained, the quantitative model of each element is mapped in 2 dimension coordinate systems, obtains the broad sense face adjacent map of assembling model.

Beneficial effect

By the present invention, assembling model is changed into GFAG figure, each Quantification of elements coordinate of set of node V (x (v) that assembling model is corresponding, y (v)) node of expression GFAG, line between two points shows between two parts and has assembly constraint, like this, GFAG had both caught the topology information of assembling model, had described quantitatively again its geological information, need not do further conversion simultaneously and can carry out the comparison of topology information and geological information.After assembling model was changed into GFAG figure, in assembling model, the excavation of universal design structure and general Graph-based data mining technology were basic identical.Because the graph-based mining algorithm is comparatively ripe, therefore can further realize the excavation of universal design structure in assembling model by the graph-based mining algorithm by the present invention.

Description of drawings

Fig. 1 is the assembling model quantitative description flow process of excavating towards the universal design structure;

Fig. 2 is the axonometric drawing of the assembling model of clamp assembly 1;

Fig. 3 is the axonometric drawing of the assembling model of clamp assembly 2;

Fig. 4 is the axonometric drawing of the assembling model of clamp assembly 3;

Fig. 5 is the axonometric drawing of the assembling model of clamp assembly 4;

Fig. 6 is that the GFAG of the assembling model of clamp assembly 1 quantitatively schemes to describe;

Fig. 7 is that the GFAG of the assembling model of clamp assembly 2 quantitatively schemes to describe;

Fig. 8 is that the GFAG of the assembling model of clamp assembly 3 quantitatively schemes to describe;

Fig. 9 is that the GFAG of the assembling model of clamp assembly 4 quantitatively schemes to describe;

Figure 10 thick line is partly the GFAG figure of universal design structure in clamp assembly 1 that excavates out;

Figure 11 thick line is partly the GFAG figure of universal design structure in clamp assembly 2 that excavates out;

Figure 12 thick line is partly the GFAG figure of universal design structure in clamp assembly 3 that excavates out;

Figure 13 thick line is partly the GFAG figure of universal design structure in clamp assembly 4 that excavates out;

The axonometric drawing of the assembling model of the universal design structure that Figure 14 excavates out.

Embodiment

Below in conjunction with specific embodiment, the present invention is described:

This embodiment shows that by 4 clamp assemblies the present invention proposes the effect of assembling model quantitative description in the universal design structure is excavated, and the axonometric drawing of 4 clamp assembled models is respectively as Fig. 2-shown in Figure 5.Particularly, this embodiment is quantitatively described 4 assembling model by following 5 steps: the set of node V and the limit collection E that one, obtain corresponding GFAG figure from assembling model, two, quantize geometric element information, three, quantize single part model, four, quantize limit collection E, five, quantize the set of node V of GFAG figure, complete the quantitative description to assembling model.Realize on this basis the excavation of universal design structure in 4 assembling model by general graph-based mining algorithm.

Concrete steps are:

Step 1: the set of node V and limit collection E, wherein the set of node V={v that generate broad sense face adjacent map (GFAG figure) according to the part in assembling model ₁, v ₂... } and in element v _iCorresponding to a part in assembling model, limit collection E={e ₁, e ₂... } and in element e _iCorresponding to the assembly restriction between certain two node, if node v _i, v _jThere is assembly restriction between corresponding part, v _iAnd v _jBetween have a limit e.

Step 2: set up the geometric element information quantization model of each geometric surface of part in assembling model, for geometric surface f, its geometric element information is determined by following formula:

$\left\{\begin{array}{c}x\left(f\right)=\underset{i=1}{\overset{n_l}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i[p_x\left(f\right)+p_{e,x}(f,f_i)][p_x\left(f_i\right)+p_{e,x}(f,f_i)]\\ y\left(f\right)=\underset{i=1}{\overset{n_l}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i[p_y\left(f\right)+p_{e,y}(f,f_i)][p_y\left(f_i\right)+p_{e,y}(f,f_i)]\end{array}\right.$

Wherein, f is current geometric surface, f _iBe i the adjacent surface of f, n _lQuantity for the adjacent surface of f; ε _i=(T (f) T (f _i)/θ (f, f _i)), T (f) means the integer of Noodles type: corresponding T (f)=1 when f is the plane, corresponding T (f)=2 when f is the face of cylinder, corresponding T (f)=3 when f is circular conical surface, corresponding T (f)=4 when f is sphere, corresponding T (f)=5 when f is anchor ring, corresponding T (f)=6 when f is free form surface; θ (f, f _i) represent face f and f _iBetween two subangles, p (f) and p _e(f, f _i) be the form parameter of describing geometric surface and geometrical edge, p _x, p _y, p _e,x, p _{E, y}The x component and the y component that represent respectively the respective shapes parameter; For this step, detailed process is at document " Ma LJ, Huang ZD, Wang YW.Automatic discovery of common design structures in CAD models.Computers﹠amp; Graphics, 2010; 34:545-555. " obtaining open explanation, these class methods have been eliminated the impact of subjective factor the largelyst.

Step 3: the geometric element information quantization model of each geometric surface of each part in the assembling model that obtains according to step 2, set up the quantitative model of single part model, because all part models all are comprised of geometric surface, therefore, adopt the mean value of the quantitative information of each geometric surface in part model to describe the quantitative information of part model:

$\left\{\begin{array}{c}x\left(p\right)=\underset{i=1}{\overset{n_f}{\mathrm{\Σ}}}x\left(f_i\right)/n_f\\ y\left(p\right)=\underset{i=1}{\overset{n_f}{\mathrm{\Σ}}}y\left(f_i\right)/n_f\end{array}\right.$

Wherein p represents part, and x (f) and y (f) are calculated by step 2, n _fSum for geometric surface in part model p; By averaging, a part can be regarded as " a broad sense face " with the equivalence of part geometry face.The part sum that comprises in clamp assembly 1-4 is respectively: 27,28,25 and 28.

Step 4: the geometric element information quantization model of each geometric surface of each part in the assembling model that obtains according to step 2, set up the quantitative model of assembly constraint between two parts; Usually, two part p in assembling model _i, p _jBetween assembly constraint can be decomposed into some fiting constraints to geometric surface, the fiting constraint type of synthetic geometry face and geometric surface form parameter are to the assembly constraint between part in assembling model here, namely in GFAG figure, the element e of limit collection E is described:

$\left\{\begin{array}{c}x\left(e\right)=x(p_i,p_j)=\underset{k=1}{\overset{n_e}{\mathrm{\Σ}}}\frac{T(f_{i,k},f_{j,k})[x\left(f_{i,k}\right)+x\left(f_{j,k}\right)]}{2n_e}\\ y\left(e\right)=y(p_i,p_j)=\underset{k=1}{\overset{n_e}{\mathrm{\Σ}}}\frac{T(f_{i,k},f_{j,k})[y\left(f_{i,k}\right)+y\left(f_{j,k}\right)]}{2n_e}\end{array}\right.$

F wherein _i,kAnd f _j,kExpression part p _i, p _jBetween the k that cooperatively interacts to geometric surface; T(f _i, f _j) be the fiting constraint type of geometric surface, " being harmonious " retrains corresponding T (f _i, f _j)=1, " contact " retrain corresponding T (f _i, f _j)=2, " skew " retrain corresponding T (f _i, f _j)=3, " angle " retrain corresponding T (f _i, f _j)=4; n _eBe two part p _i, p _jBetween the logarithm of the geometric surface that cooperatively interacts; X (f) and y (f) are calculated by step 2;

Step 5: the quantitative model of assembly constraint between the quantitative model of the single part model that obtains according to step 3 and two parts that step 4 obtains, set up the quantitative model of each element in set of node V:

$\left\{\begin{array}{c}x\left(v\right)=\underset{i=1}{\overset{n_p}{\mathrm{\Σ}}}[x\left(p\right)+x(p,p_i)][x\left(p_i\right)+x(p,p_i)]\\ y\left(v\right)=\underset{i=1}{\overset{n_p}{\mathrm{\Σ}}}[y\left(p\right)+y(p,p_i)][y\left(p_i\right)+y(p,p_i)]\end{array}\right.$

Wherein p is current part model, p _iBe i the part model that coordinates with current part, n _pQuantity for mating parts; X (p) and y (p) are calculated by step 3; X (p, p _i) and y (p, p _i) calculated by step 4;

Step 6: in the set of node V that step 5 is obtained, the quantitative model of each element is mapped in 2 dimension coordinate systems, coordinate in this system (x (v), y (v)) node of expression GFAG, corresponding to a part in assembling model, line between two points shows between two parts and has assembly constraint, obtains the broad sense face adjacent map of assembling model.

Like this, GFAG had both caught the topology information of assembling model, had described quantitatively again its geological information, need not do further conversion simultaneously and can carry out the comparison of topology information and geological information.Fig. 6-Fig. 9 is respectively that the GFAG figure of Fig. 2-clamp assembled model shown in Figure 5 describes.

After the assembling model of clamp assembly 1-4 is converted into GFAG figure, just can excavate out 4 universal design structures in assembling model by general graph-based mining algorithm.This embodiment adopts AGM (Apriori-based Graph Mining) algorithm to carry out the excavation of universal design structure.Universal design structure in the assembling model of excavation card release board component 1-4 is as shown in the part of the thick line in Figure 10-Figure 13.Compare by the assembling model with clamp assembly 1-4, finally determine the model of universal design structure as shown in figure 14.

This embodiment shows, the assembling model quantitative description that the present invention proposes can be used for the excavation of universal design structure, but its universal design structure of excavating out may be less than the scope of reality expectation, and this is the follow-up improved direction of the present invention.

Claims (1)

1. assembling model quantitative description of excavating towards the universal design structure is characterized in that: adopt following steps:

Step 1: the set of node V and limit collection E, wherein the set of node V={v that generate broad sense face adjacent map according to the part in assembling model ₁, v ₂... } and in element v _iCorresponding to a part in assembling model, limit collection E={e ₁, e ₂... } and in element e _iCorresponding to the assembly restriction between certain two node;

Step 2: set up the geometric element information quantization model of each geometric surface of part in assembling model, for geometric surface f, its geometric element information is determined by following formula:

$\left\{\begin{array}{c}x\left(f\right)=\underset{i=1}{\overset{n_l}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i[p_x\left(f\right)+p_{e,x}(f,f_i)][p_x\left(f_i\right)+p_{e,x}(f,f_i)]\\ y\left(f\right)=\underset{i=1}{\overset{n_l}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i[p_y\left(f\right)+p_{e,y}(f,f_i)][p_y\left(f_i\right)+p_{e,y}(f,f_i)]\end{array}\right.$

Wherein, f is current geometric surface, f _iBe i the adjacent surface of f, n _lQuantity for the adjacent surface of f; ε _i=(T (f) T (f _i)/θ (f, f _i)), T (f) means the integer of Noodles type: corresponding T (f)=1 when f is the plane, corresponding T (f)=2 when f is the face of cylinder, corresponding T (f)=3 when f is circular conical surface, corresponding T (f)=4 when f is sphere, corresponding T (f)=5 when f is anchor ring, corresponding T (f)=6 when f is free form surface; θ (f, f _i) represent face f and f _iBetween two subangles, p (f) and p _e(f, f _i) be the form parameter of describing geometric surface and geometrical edge, p _x, p _y, p _e,x, p _{E, y}The x component and the y component that represent respectively the respective shapes parameter;

Step 3: the geometric element information quantization model of each geometric surface of each part in the assembling model that obtains according to step 2, set up the quantitative model of single part model:

$\left\{\begin{array}{c}x\left(p\right)=\underset{i=1}{\overset{n_f}{\mathrm{\Σ}}}x\left(f_i\right)/n_f\\ y\left(p\right)=\underset{i=1}{\overset{n_f}{\mathrm{\Σ}}}y\left(f_i\right)/n_f\end{array}\right.$

Wherein p represents part, and x (f) and y (f) are calculated by step 2, n _fSum for geometric surface in part model p;

Step 4: the geometric element information quantization mould of each geometric surface of each part in the assembling model that obtains according to step 2

Type, set up the quantitative model of assembly constraint between two parts:

$\left\{\begin{array}{c}x\left(e\right)=x(p_i,p_j)=\underset{k=1}{\overset{n_e}{\mathrm{\Σ}}}\frac{T(f_{i,k},f_{j,k})[x\left(f_{i,k}\right)+x\left(f_{j,k}\right)]}{2n_e}\\ y\left(e\right)=y(p_i,p_j)=\underset{k=1}{\overset{n_e}{\mathrm{\Σ}}}\frac{T(f_{i,k},f_{j,k})[y\left(f_{i,k}\right)+y\left(f_{j,k}\right)]}{2n_e}\end{array}\right.$

F wherein _i,kAnd f _j,kExpression part p _i, p _jBetween the k that cooperatively interacts to geometric surface; T(f _i, f _j) be the fiting constraint type of geometric surface, " being harmonious " retrains corresponding T (f _i, f _j)=1, " contact " retrain corresponding T (f _i, f _j)=2, " skew " retrain corresponding T (f _i, f _j)=3, " angle " retrain corresponding T (f _i, f _j)=4; n _eBe two part p _i, p _jBetween the logarithm of the geometric surface that cooperatively interacts; X (f) and y (f) are calculated by step 2;

Step 5: the quantitative model of assembly constraint between the quantitative model of the single part model that obtains according to step 3 and two parts that step 4 obtains, set up the quantitative model of each element in set of node V:

$\left\{\begin{array}{c}x\left(v\right)=\underset{i=1}{\overset{n_p}{\mathrm{\Σ}}}[x\left(p\right)+x(p,p_i)][x\left(p_i\right)+x(p,p_i)]\\ y\left(v\right)=\underset{i=1}{\overset{n_p}{\mathrm{\Σ}}}[y\left(p\right)+y(p,p_i)][y\left(p_i\right)+y(p,p_i)]\end{array}\right.$

Wherein p is current part model, p _iBe i the part model that coordinates with current part, n _pQuantity for mating parts; X (p) and y (p) are calculated by step 3; X (p, p _i) and y (p, p _i) calculated by step 4;

Step 6: in the set of node V that step 5 is obtained, the quantitative model of each element is mapped in 2 dimension coordinate systems, obtains the broad sense face adjacent map of assembling model.

Images