CN107944015A - Threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity - Google Patents
Threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity Download PDFInfo
- Publication number
- CN107944015A CN107944015A CN201711308002.7A CN201711308002A CN107944015A CN 107944015 A CN107944015 A CN 107944015A CN 201711308002 A CN201711308002 A CN 201711308002A CN 107944015 A CN107944015 A CN 107944015A
- Authority
- CN
- China
- Prior art keywords
- face
- model
- node
- typical structure
- simulated annealing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
Abstract
The present invention is excavated for the threedimensional model typical structure based on simulated annealing and method for evaluating similarity, inventive algorithm is analyzed from the Result of typical structure, and output result is used as using Maximum Clique, and the excavation characteristic of this " fuzzy pattern " can return to the model of multi-level similarity, species is enriched has very big flexibility comprehensively, facilitates designer to carry out the design work of different phase.Since algorithm ensures that maximum public subgraph is connection, the Result of typical structure more meets the similitude perception of people.Simulated annealing describes the simple using flexible of problem, is subject to primary condition constraint few, and random factor is introduced compared with traditional search strategy makes it be more suitable for solving maximal clique problem.Shown by the statistical test of model library, the typical structure based on simulated annealing the method for the present invention excavates performance will be apparently higher than other algorithms of this area.The time efficiency of the method for the present invention is higher than this area other methods.
Description
Technical field
The present invention relates to a kind of design reuse method of engineering goods three-dimensional CAD model, particularly one kind to be moved back based on simulation
The threedimensional model typical structure of fire excavates and method for evaluating similarity.
Technical background
With deepening continuously for Products Digital technology application, the largely abundant CAD design achievement that enterprise has accumulated is
The development of new product provides the reusing design resource of preciousness.Although product be continuously updated regenerate, brand-new function,
Structure and technological design only have about 20%, remaining 80% design can have then been designed come complete by directly reusing or locally changing
Into product has very strong similitude and inheritance in function, structure and technique etc..In the model library of enterprise, design
Personnel often define design reuse of some typical structures for model according to different application backgrounds by man-machine interactively mode.
The model library huge in face of enterprise, this method inefficiency, is influenced greatly by subjective factor, and substantial amounts of typical structure is often
Lie in the entirely different CAD model of shape, designer is difficult to find.Therefore, it is how fast from the product model of magnanimity
Speed, effectively excavate the typical structure of needs, and one reused to it as each link of product development compels to be essential
Ask.
So far, existing a variety of Design of digital method for reusing are used widely in Design of Mechanical Product, such as base
Design in case-based reasoning, design based on modular design, based on retrieval etc., can effectively improve product design efficiency and matter
Amount, causes the pay attention to day by day of numerous scientific research personnel.The design of case-based reasoningf is to pass by similar ask in knowledge base by accessing
The solution of topic is so as to obtain a kind of reasoning pattern that current problem solves method.Based on it is modular design be to difference in functionality,
The product of performance and specification is divided and designs a series of function module, is made up of not the selection and combination of module
Same product.Design based on retrieval mainly by establishing 3 d part CAD model storehouse, designs certain matching algorithm, in mould
Found in type storehouse and be designed reuse with the highest existing parts of design requirement similarity.
Since engineering goods CAD model includes more geometry and topology information, case-based reasoningf and based on modular
Design reuse method reliability is not high, simultaneously because in modern designs three-dimensional CAD model design information complication, it is traditional
Design of digital method for reusing based on retrieval is no longer applicable in, and heuritic approach is due to can carry out large amount of complex information
Quick and precisely there are more and more applications in Digitized Design for Mechanical Product field the characteristics of processing.
On the whole, existing Design of digital reuse technology is a large amount of in product development mainly for product component level
Design reuse be then that similarity determination is with qualitative and text attribute on the parts internal feature and typical structure more carefully seen
Based on matching, and the experience and knowledge for determining nor being often relied on people of reusing design information, Product Digitalization Design reuse
Coarse size, low precision, intelligent level it is low.
The content of the invention
In order to solve the above technical problem, the present invention provides a kind of threedimensional model typical structure digging based on simulated annealing
Pick and method for evaluating similarity, its technical solution are:Comprise the following steps:
Step 1:The three-dimensional CAD model of STEP document presentations is read in into the three-dimensional CAD that Open CASCADE are geometric modeling core
The B-Rep information on model index system, extraction model face and side;
Step 2:The face node of three-dimensional CAD model and the attribute on side reflected according to Model B-Rep information, builds three-dimensional CAD
The attribute adjacent map of model;
Step 3:Using with vertex and side between the local typical structure and three-dimensional CAD model attribute adjacent map for reusing value
Mapping relations establish associated diagram and its associated diagram matrix;
Step 4:Clique in associated diagram is detected based on simulated annealing, realizes that the typical structure of three-dimensional CAD model is dug
Pick;
Step 5:According to typical structure Result, matching typical structure similarity is calculated.
Further, in step 2, for the attribute adjacent map of CAD model, each face f in modeliHave unique
Node ViCorrespond, the attribute set W in face in modelVThe direction of type, face including face, the relative area in face;E is face
Between syntople, for any two face f in modeliAnd fjIf having syntople between two faces, have unique
A line EijCorrespond to therewith, the attribute set W on side in modelEAngle, common edge between type, adjacent surface including side it is recessed
Convexity.
Further, the type in face:By the face of CAD model be divided into plane class, cylinder noodles, annulus noodles, circular cone noodles,
Spherical class and free form surface class;
The direction in face:The relation being defined as in geometry kernel to the direction in CAD model face between the direction in face and surface direction;
Surface direction is from center or is axially directed to outside, and the direction in face is directed toward outside from model entity;Face to face with the direction phase of curved surface
It is false when different meanwhile face is oriented to true;
The relative area in face:In CAD model, some face fiArea be Si, its all of its neighbor face collection is combined into FA={f1,
f2,…,fk, then the relative area S in the faceeiFor:
The type on side:The side of CAD model is divided into straight line, circular curve, elliptic curve, hyperbola, parabola and other songs
Line etc..
Angle between adjacent surface:Point three kinds of situations, 1. two adjacent junctions are all plane, and the angle between face is with two
The normal direction angle calcu-lation in face;2. a face is plane, another face is quadratic surface, and plane is followed the example of to quadratic surface takes axial direction;
3. two faces are all quadratic surface, the angle calcu-lation of two curved surfaces of angle between face;Do not considered for irregular curved surface,
Then the adjacent surface angle theta on side is:
The concavity and convexity of common edge:The concavity and convexity on side is divided into chimb, concave edge peace slide.
Further, the concrete operations of step 2 are as follows:
Step 2.1:Each face of B-Rep models is traveled through, which is added to the face set F of modelS={fi, 1≤i≤m, its
Middle m is the number in face in model;
Step 2.2:Ergodic surface set FSIn each face fi, while a figure corresponding with the face is created in attribute adjacent map
Node, and extract the attribute of the attribute as its corresponding node in the face;
Step 2.3:For FSIn each two face fiWith fj, the connection relation between them is calculated, if two sides is adjacent,
Build a line between corresponding two node in attribute adjacent map, the attribute on the side by the corner dimension between face, serial relation and
The concavity and convexity on side determines.
Further, in step 3, by the detection of public subgraph between typical structure and three-dimensional CAD model attribute adjacent map
Problem, the Mining Problems of Clique are changed into by building associated diagram and associated diagram matrix, give two attribute adjacent map G1, G2,
Set of node V1, V2If the associated diagram of the two is HV, then the specific algorithm step of structure associated diagram is as follows:
Step 3.1:To scheming G1In any one node vi∈V1(1≤i≤n), traversing graph G2In each node uj∈V2(1
≤ j≤m), composition node is to (vi, uj), if viWith ujWith identical property value, then by (vi, uj) add associated diagram HVSection
Point set VHAs one node.
Step 3.2:Appoint and take associated diagram HVIn two node uH=(u1, u2) and vH=(v1, v2), if u1≠v1, u2≠v2, and
Scheme G1In side e1=(u1, v1) with scheming G2In side e2=(u2, v2) there is identical property value or figure G1In node u1And v1, figure
G2In node u2And v2It is each non-conterminous, then construct a line eH=<uH, vH>, and it is added to associated diagram HVSide collection EHIt is middle to be used as it
A line, wherein, the first type when being known as connecting, second type when being known as non-interconnected.
Further, in step 4, the feasible solution of maximal clique problem is a vertex set for meeting certain constraints,
The matrix G of the associated diagram of canonical correlation structural information in CAD model is retrieved as input(Matrix size is n);The son excavated
Maximum Clique S in structure matrix, that is, associated diagram matrix is output;Simulated annealing comprises the following steps that:
Step 4.1:S is utilized according to subgraph size kwapFunction randomly selects node and forms initial subgraph S from containerinitial, together
Shi Liyong object function function Fit (S)=k* (k-1)-MijCalculate the fitness f of initial subgraphinitial, remaining n-k nodes composition
Remaining subgraph Sremainder;
Step 4.2:Into circulation, with reference to SwapFunction is any from current initial subgraph to choose node u, utilizes Cald function meters
U is calculated in SinitialThe degree d on middle vertexu, from remaining subgraph SremainderIn arbitrarily choose a node v, calculate v replace u after exist
SinitialThe degree d on middle vertexvIf dvValue is not less than du, just receive new construction, and if dvLess than du, from remaining subgraph
SremainderIn again arbitrarily choose node, and if repeat 8n time afterwards dvStill less than du, then new construction degree be unsatisfactory for requiring;
Step 4.3:Calculate the adaptive value f of new constructionnewIf finitial<=fnew, receive this structure, if finitial>fnew,
This structure is then received with certain probability;
Step 4.4:Algorithm temperature T is successively decreased with given pace r, and circulation performs step Step2, until searching out target adaptive value f
Structure or temperature reach terminating point TendAfterwards, program exits, that is, completes the excavation of this typical structure.
Further, in step 5, searched using simulated annealing in associated diagram during Clique, find pole
Agglomerate, the Maximum Clique found out corresponds to the public subgraph that typical structure and CAD model are jointly comprised, using the section of Maximum Clique
Point number divided by the number in typical structure face describe local similarity size, i.e.,
Wherein, S represents local similarity value, NmRepresent the number of all Maximum Clique nodes, N represents in face in typical structure
Number, S is bigger to represent that two model local similarities are higher, that is, obtains retrieval result.
Beneficial effects of the present invention are:Compared with prior art, technique effect of the invention includes:
(1)Analyzed from the Result of typical structure, inventive algorithm is using Maximum Clique as output as a result, and this " fuzzy mould
The excavation characteristic of formula " can return to the model of multi-level similarity, and species is enriched has very big flexibility comprehensively, convenient design
Personnel carry out the design work of different phase.And since inventive algorithm ensures that maximum public subgraph is to connect, Dian Xingjie
The similitude that the Result of structure more meets people perceives.
(2)Since simulated annealing describes that problem is simple, using flexible, it is subject to primary condition constraint few, is searched with tradition
Rope strategy makes it be more suitable for solving maximal clique problem compared to random factor is introduced.Pass through the statistical test table of model library
Bright, the typical structure based on simulated annealing the method for the present invention excavates performance will be apparently higher than other algorithms of this area.
(3)It is Intel Pentium 4 CPU 3.06GHz, memory 4GB to test used test machine CPU.Inventive algorithm
Time complexity be concentrated mainly on and search Clique using simulated annealing, single model average handling time is
0.136s.The time efficiency of the method for the present invention is higher than this area other methods.
Brief description of the drawings
Fig. 1 is flow chart of the method for the present invention.
Fig. 2 is the direction schematic diagram in CAD model face.
Fig. 3 is the concavity and convexity schematic diagram on CAD model side.
Fig. 4(a)For typical structure;
Fig. 4(b)For Fig. 4(a)Typical structure CAD model attribute adjacent map.
Fig. 5(a)For the associated diagram of two attribute adjacent map structures in Fig. 4;
Fig. 5(b)For Fig. 5(a)Associated diagram matrix.
Fig. 6 is based on simulated annealing canonical correlation structure retrieval result for the present invention.
Fig. 7 is that the algorithm performance that typical structure of the present invention excavates compares.
Fig. 8(a)Complete-precision ratio curve map is looked into search being averaged for ESB model libraries using the present invention;
Fig. 8(b)Complete-precision ratio curve map is looked into search being averaged for talents for agri-mechanization model library using the present invention.
Embodiment
The present invention proposes a kind of threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity, this hair
Bright method is using the Clique that associated diagram is constructed between simulated annealing detection CAD model attribute adjacent map, effectively to dig
Pick model in typical structure and complete similarity evaluation, the method for the present invention can effectively support the three-dimensional CAD of design field
The reuse of aspect of model level and local structural level design information.
As shown in Figure 1, the threedimensional model typical structure proposed by the present invention based on simulated annealing excavates and similarity evaluation
Method comprises the following steps:
Step 1:The three-dimensional CAD model of STEP document presentations is read in into the three-dimensional CAD that OpenCASCADE is geometric modeling core
The B-Rep information on model index system, extraction model face and side;
Step 2:The face node of three-dimensional CAD model and the attribute on side reflected according to Model B-Rep information, builds three-dimensional CAD
The attribute adjacent map of model;
Step 3:Using with vertex and side between the local typical structure and three-dimensional CAD model attribute adjacent map for reusing value
Mapping relations establish associated diagram and its associated diagram matrix;
Step 4:Clique in associated diagram is detected based on simulated annealing, realizes that the typical structure of three-dimensional CAD model is dug
Pick;
Step 5:According to typical structure Result, matching typical structure similarity is calculated.
In step 1 of the present invention, for the attribute adjacent map of CAD model, each face in modelf i There is unique nodeV i Correspond, the attribute set in face in modelW V The direction of type, face including face, the relative area in face etc..EFor face it
Between syntople, for any two face in modelf i Withf j If having syntople between two faces, have unique
A lineE ij Correspond to therewith, the attribute set on side in modelW E Angle, common edge between type, adjacent surface including side it is recessed
Convexity etc..
The type in face:The face of CAD model is divided into plane class, cylinder noodles, annulus noodles, circular cone noodles, ball by the present invention
Noodles and free form surface class etc..
The direction in face:The direction in CAD model face is defined as between the direction in face and surface direction in geometry kernel
Relation, the direction on surface always from center or are axially directed to outer, and the direction in face is always outside model entity direction.Face to face
When identical with the direction of curved surface, face is oriented to true, is false when different.As shown in Fig. 2, cylindrical surfacef 1 Be oriented to
True, cylindrical holef 2 Direction be then false.
The relative area in face:In CAD model, some facef i Area beS i , its all of its neighbor face collection is combined intoF A ={f 1 ,f 2 ,…,f k , then the relative area in the faceS ei For:
The type on side:The present invention by the side of CAD model be divided into straight line, circular curve, elliptic curve, hyperbola, parabola and
Other curves etc..
Adjacent surface angle:The present invention considers following three kinds of situations, and two adjacent junctions all be plane, the angle use between face
The normal direction angle calcu-lation in two faces;One face is plane, another face is quadratic surface, and plane is followed the example of to quadratic surface takes axis
To;Two faces are all quadratic surface, the angle calcu-lation of two curved surfaces of angle between face.Not examined for irregular curved surface
Consider, then the adjacent surface angle on sideθFor:
The concavity and convexity on side:The concavity and convexity on side is divided into chimb, concave edge peace slide by the present invention.As shown in figure 3, to the recessed of model side
Convexity judging result is:Sidel 1 For chimb,l 2 For smooth side,l 3 For concave edge.
In step 2 of the present invention, by extracting the B-Rep information in three-dimensional CAD model, by CAD model structure attribute adjoining
The specific steps of figure can be described as follows:
Step 2.1:Each face of B-Rep models is traveled through, the face which is added to model is gatheredFS={f i , 1≤i≤m, its
InmFor the number in face in model;
Step 2.2:Ergodic surface setFSIn each facef i , while a figure corresponding with the face is created in attribute adjacent map
Node, and extract the attribute of the attribute as its corresponding node in the face;
Step 2.3:ForFSIn each two facef i Withf j , the connection relation between them is calculated, if two sides is adjacent,
Build a line between corresponding two node in attribute adjacent map, the attribute on the side by the corner dimension between face, serial relation and
The concavity and convexity on side determines.
It is illustrated in figure 4 typical structure and comprising the typical structure CAD model attribute adjacent map, the attributed graph of CAD model
Model geometric and topology information are have recorded, what is covered contains much information, and message form is succinct, and expression is directly perceived, readable strong.
In step 3 of the present invention, based on step 2 of the present invention, by typical structure and three-dimensional CAD model attribute adjacent map it
Between public subgraph test problems, change into the Mining Problems of Clique by building associated diagram and associated diagram matrix, give
Two attribute adjacent mapsG 1 ,G 2 , set of nodeV 1 ,V 2 If the associated diagram of the two isH V , then the specific algorithm step of associated diagram is built such as
Under:
Step 3.1:To figureG 1 In any one nodev i ∈V 1 (1≤i≤n), traversing graphG 2 In each nodeu j ∈V 2 (1
≤j≤m), composition node to (v i ,u j ), ifv i Withu j With identical property value, then will (v i ,u j ) add associated diagramH V Section
Point setV H As one node.
Step 3-2:Appoint and take associated diagramH V In two nodesu H =(u 1 ,u 2 ) andv H =(v 1 ,v 2 ), ifu 1 ≠v 1 ,u 2 ≠v 2 , and
FigureG 1 In sidee 1 =(u 1 ,v 1 ) with schemingG 2 In sidee 2 =(u 2 ,v 2 ) there is identical property value or figureG 1 In nodeu 1 Withv 1 , figureG 2 In nodeu 2 Withv 2 It is each non-conterminous, then construct a linee H =<u H ,v H >, and it is added to associated diagramH V Side collectionE H It is middle to be used as it
A line, wherein, the first type when being known as connecting, second type when being known as non-interconnected.
Such as Fig. 5(a)It show the associated diagram built according to two attribute adjacent maps in Fig. 4, digitized representation node of graph in figure
Label, according to step1, searches the identical vertex of node type to set, associated diagram is by node NaA、NaB、NaG、NbD、NbF、NdA、
NdB、NdG、NcCAnd NcEComposition;According to step2,10 vertex of associated diagram are connected to being attached according to the attribute on each node side
Logical side is indicated by the solid line, and non-interconnected side is represented by dashed line, and the vertex set in dotted rectangle is the Clique of associated diagram, office
Portion's structure Mining Problems are converted to maximal clique problem in detection associated diagram.For the convenience of Solve problems, associated diagram is built
Adjacency matrix, such as Fig. 5(b)Shown, numeral 0 does not have connected relation, numeral 1 to be between representing node between representing node in matrix
Non-interconnected relation, it is connected relation between the two that numeral 2, which represents,.
In step 4 of the present invention, the maximal clique problem of associated diagram is solved based on simulated annealing.Maximal clique problem can
Row solution is a vertex set for meeting certain constraints, in order to which the expression with solution is consistent, improves efficiency of algorithm, the present invention
So as to retrieve the matrix of the associated diagram of canonical correlation structural information in CAD modelGAs input(Matrix size isn);At the same time
In order to strengthen the diversity for returning to model easily and flexibly to carry out the work of the design reuse of different design stage, with what is excavated
Maximum Clique in minor structure matrix, that is, associated diagram matrixSFor output(Minor structure to ensure excavated possesses certain engineering significance,
Provide minor structure matrix sizekMinimum value is 3).In simulated annealing of the present invention, initial temperatureTIt is set as 1000, terminates
TemperatureT end It is set as 0.1, temperature varying coefficientrIt is 0.99, object function isFit(S)=k*(k-1)-M ij To calculate subgraph
Fitnessf, in functionM ij It is not the sum of 0 number to solve element in certain Clique, when searching out target Clique, function
Adaptive value is 0.Simulated annealing comprises the following steps that:
Step 4.1:According to subgraph sizekUtilizeSwapFunction randomly selects node and forms initial subgraph from containerS initial ,
Utilize object function function at the same timeFit(S)=k*(k-1)-M ij Calculate the fitness of initial subgraphf initial , it is remainingn-kNode structure
Into remaining subgraphS remainder ;
Step 4.2:Into circulation, with reference toSwapFunction is any from current initial subgraph to choose nodeu, utilizeCaldFunction meter
Calculateu S initial The degree on middle vertexd u , from remaining subgraphS remainder In arbitrarily choose a nodev, calculatevReplaceuExist afterwardsS initial The degree on middle vertexd v Ifd v Value is not less thand u , just receive new construction, and ifd v It is less thand u , from remaining subgraphS remainder In again arbitrarily choose node, and if repeat 8nAfter secondaryd v Still less thand u , then new construction degree be unsatisfactory for requiring;
Step 4.3:Calculate the adaptive value of new constructionf new Iff initial <=f new , receive this structure, iff initial >f new ,
This structure is then received with certain probability;
Step 4.4:Algorithm temperatureTWith given pacerSuccessively decrease, circulation performs step Step2, until searching out target adaptive valuef
Structure or temperature reach terminating pointT end Afterwards, program exits, and completes the excavation of this typical structure.
In step 5 of the present invention, searched using simulated annealing in associated diagram during Clique, it is only necessary to search
To Maximum Clique, the Maximum Clique found out corresponds to the public subgraph that typical structure and CAD model are jointly comprised, therefore adopts
Local similarity size is described with the node number of Maximum Clique divided by the number in typical structure face.
Wherein,SRepresent local similarity value,N m Represent the number of all Maximum Clique nodes,NRepresent face in typical structure
Number,STwo model local similarities of bigger expression are higher.
Experiment 1
The typical structure for being illustrated in figure 6 the selection dark signs from model excavates it, and obtained in model library
Return to model.It can be found that the model with the partial structurtes can be all retrieved by the method for the present invention from model library,
Some essentially identical models are eliminated, 15 representative models is have chosen and is shown.Can in model from returning
To find out, canonical correlation structure is often hidden in the different three-dimensional CAD model of shape, and inventive algorithm can be effective
The CAD model of implicit typical structure is pushed to designer by ground, so that designer refers to.
Experiment 2
As shown in fig. 7, inventive algorithm and ant group algorithm and genetic algorithm be in general field model library, using disk-shaped structure for survey
The typical structure mining ability of examination object compares, and after removing independence model, will return to model and is ranked up by similarity size.From
Result is analyzed, and relative to ant group algorithm, inventive algorithm is using Maximum Clique as exporting as a result, and this " fuzzy pattern "
The model of multi-level similarity can be returned to by excavating characteristic, and since inventive algorithm ensures that maximum public subgraph is connection
, the Result of typical structure more meets the similitude perception of people, such as Fig. 7 ant group algorithms the 2nd and the 6th return to model and are not inconsistent
Close visual similarity;Relative to genetic algorithm, the return model quantity of inventive algorithm is more, and has higher complexity,
Depth excavation is carried out easy to designer.
Experiment 3
In order to fully contrast the performance of three kinds of algorithms, the present invention is respectively to general field ESB model libraries and talents for agri-mechanization mould
CAD model in type storehouse carries out statistical test, obtains recall level average-precision ratio(Precision-Recall, PR)Curve,
The curve of preferable retrieval result should be the parallel lines that a precision ratio is constantly equal to 1.0, and the top curve in position has higher
Precision, represents preferable retrieval result.From PR curves it will be clear that the retrieval performance of inventive algorithm is obvious high
In ant group algorithm and genetic algorithm, as shown in Figure 8.
The above is to the preferred embodiment of the present invention, it should be noted that for the ordinary skill people of the art
For member, without departing from the principle of the present invention, some improvement and increase can also be made, these are improved and increase also regards
For the protection to the present invention.
Claims (7)
1. the threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity, it is characterised in that including following
Step:
Step 1:The three-dimensional CAD model of STEP document presentations is read in into the three-dimensional CAD that Open CASCADE are geometric modeling core
The B-Rep information on model index system, extraction model face and side;
Step 2:The face node of three-dimensional CAD model and the attribute on side reflected according to Model B-Rep information, builds three-dimensional CAD
The attribute adjacent map of model;
Step 3:Using with vertex and side between the local typical structure and three-dimensional CAD model attribute adjacent map for reusing value
Mapping relations establish associated diagram and its associated diagram matrix;
Step 4:Clique in associated diagram is detected based on simulated annealing, realizes that the typical structure of three-dimensional CAD model is dug
Pick;
Step 5:According to typical structure Result, matching typical structure similarity is calculated.
2. the threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity as claimed in claim 1, its
It is characterized in that, in step 2, for the attribute adjacent map of CAD model, each face f in modeliThere is unique node ViTherewith
It is corresponding, the attribute set W in face in modelVThe direction of type, face including face, the relative area in face;Adjoinings of the E between face
Relation, for any two face f in modeliAnd fjIf having syntople between two faces, there is unique a line EijWith
Correspondence, the attribute set W on side in modelEAngle, the concavity and convexity of common edge between type, adjacent surface including side.
3. the threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity as claimed in claim 2, its
It is characterized in that,
The type in face:The face of CAD model is divided into plane class, cylinder noodles, annulus noodles, circular cone noodles, spherical class and freedom
Curved surface class;
The direction in face:The relation being defined as in geometry kernel to the direction in CAD model face between the direction in face and surface direction;
Surface direction is from center or is axially directed to outside, and the direction in face is directed toward outside from model entity;Face to face with the direction phase of curved surface
It is false when different meanwhile face is oriented to true;
The relative area in face:In CAD model, some face fiArea be Si, its all of its neighbor face collection is combined into FA={f1,f2,…,
fk, then the relative area S in the faceeiFor:
The type on side:The side of CAD model is divided into straight line, circular curve, elliptic curve, hyperbola, parabola and other songs
Line, the angle between adjacent surface:Point three kinds of situations, 1. two adjacent junctions are all plane, two faces of the angle between face
Normal direction angle calcu-lation;2. a face is plane, another face is quadratic surface, and plane is followed the example of to quadratic surface takes axial direction;3. two
A face is all quadratic surface, the angle calcu-lation of two curved surfaces of angle between face;Then the adjacent surface angle theta on side is:
The concavity and convexity of common edge:The concavity and convexity on side is divided into chimb, concave edge peace slide.
4. the threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity as claimed in claim 1, its
It is characterized in that, the concrete operations of step 2 are as follows:
Step 2.1:Each face of B-Rep models is traveled through, which is added to face set FS={ f of modeli, 1≤i≤m, wherein
M is the number in face in model;
Step 2.2:Ergodic surface set FSIn each face fi, while a figure corresponding with the face is created in attribute adjacent map
Node, and extract the attribute of the attribute as its corresponding node in the face;
Step 2.3:For FSIn each two face fiWith fj, the connection relation between them is calculated, if two sides is adjacent, is being belonged to
In property adjacent map a line is built between corresponding two node.
5. the threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity as claimed in claim 1, its
It is characterized in that, in step 3, the test problems of public subgraph between typical structure and three-dimensional CAD model attribute adjacent map leads to
Cross structure associated diagram and associated diagram matrix changes into the Mining Problems of Clique, give two attribute adjacent map G1, G2, set of node
V1, V2If the associated diagram of the two is HV, then the specific algorithm step of structure associated diagram is as follows:
Step 3.1:To scheming G1In any one node vi∈V1(1≤i≤n), traversing graph G2In each node uj∈V2(1
≤ j≤m), composition node is to (vi, uj), if viWith ujWith identical property value, then by (vi, uj) add associated diagram HVSection
Point set VHAs one node;
Step 3.2:Appoint and take associated diagram HVIn two node uH=(u1, u2) and vH=(v1, v2), if u1≠v1, u2≠v2, and scheme G1
In side e1=(u1, v1) with scheming G2In side e2=(u2, v2) there is identical property value or figure G1In node u1And v1, scheme G2In
Node u2And v2It is each non-conterminous, then construct a line eH=<uH, vH>, and it is added to associated diagram HVSide collection EHIt is middle to be used as one
Bar side, wherein, the first type when being known as connecting, second type when being known as non-interconnected.
6. the threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity as claimed in claim 1, its
It is characterized in that, in step 4, the feasible solution of maximal clique problem is a vertex set for meeting certain constraints, retrieves CAD
The matrix G of the associated diagram of canonical correlation structural information is as input in model(Matrix size is n);The minor structure matrix excavated
Maximum Clique S i.e. in associated diagram matrix is output;Simulated annealing comprises the following steps that:
Step 4.1:S is utilized according to subgraph size kwapFunction randomly selects node and forms initial subgraph S from containerinitial, together
Shi Liyong object function function Fit (S)=k* (k-1)-MijCalculate the fitness f of initial subgraphinitial, remaining n-k nodes composition
Remaining subgraph Sremainder;
Step 4.2:Into circulation, with reference to SwapFunction is any from current initial subgraph to choose node u, utilizes Cald function meters
U is calculated in SinitialThe degree d on middle vertexu, from remaining subgraph SremainderIn arbitrarily choose a node v, calculate v replace u after exist
SinitialThe degree d on middle vertexvIf dvValue is not less than du, just receive new construction, and if dvLess than du, from remaining subgraph
SremainderIn again arbitrarily choose node, and if repeat 8n time afterwards dvStill less than du, then new construction degree be unsatisfactory for requiring;
Step 4.3:Calculate the adaptive value f of new constructionnewIf finitial<=fnew, receive this structure, if finitial>fnew, then
This structure is received with certain probability;
Step 4.4:Algorithm temperature T is successively decreased with given pace r, and circulation performs step Step2, until searching out target adaptive value f
Structure or temperature reach terminating point TendAfterwards, program exits, that is, completes the excavation of this typical structure.
7. the threedimensional model typical structure excavation based on simulated annealing and method for evaluating similarity as claimed in claim 1, its
It is characterized in that, in step 5, is searched using simulated annealing in associated diagram during Clique, find Maximum Clique, look into
The Maximum Clique found out corresponds to the public subgraph that typical structure and CAD model are jointly comprised, using the node number of Maximum Clique
Divided by the number in typical structure face describes local similarity size, i.e.,
Wherein, S represents local similarity value, NmRepresenting the number of all Maximum Clique nodes, N represents the number in face in typical structure,
S is bigger to represent that two model local similarities are higher, that is, obtains retrieval result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711308002.7A CN107944015A (en) | 2017-12-11 | 2017-12-11 | Threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711308002.7A CN107944015A (en) | 2017-12-11 | 2017-12-11 | Threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107944015A true CN107944015A (en) | 2018-04-20 |
Family
ID=61946482
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711308002.7A Pending CN107944015A (en) | 2017-12-11 | 2017-12-11 | Threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107944015A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110795797A (en) * | 2019-09-26 | 2020-02-14 | 北京航空航天大学 | MBD model processing feature recognition and information extraction method |
CN113806891A (en) * | 2021-09-23 | 2021-12-17 | 山东大学深圳研究院 | Rapid design method of clamp for adapting to workpiece change |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101834780A (en) * | 2010-01-28 | 2010-09-15 | 武汉理工大学 | Method for optimizing topological structure and mapping of network on chip |
CN102629287A (en) * | 2012-02-29 | 2012-08-08 | 沈阳理工大学 | Automatic identification method based on standard for the exchange of product model data-compliant numerical control data interface (STEP-NC) intersection features |
CN103093011A (en) * | 2011-11-02 | 2013-05-08 | 同济大学 | CAD (computer aided design) model based feature recognition algorithm |
US20140324904A1 (en) * | 2011-11-29 | 2014-10-30 | Hitachi, Ltd. | Similar design structure search device and similar design structure search method |
-
2017
- 2017-12-11 CN CN201711308002.7A patent/CN107944015A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101834780A (en) * | 2010-01-28 | 2010-09-15 | 武汉理工大学 | Method for optimizing topological structure and mapping of network on chip |
CN103093011A (en) * | 2011-11-02 | 2013-05-08 | 同济大学 | CAD (computer aided design) model based feature recognition algorithm |
US20140324904A1 (en) * | 2011-11-29 | 2014-10-30 | Hitachi, Ltd. | Similar design structure search device and similar design structure search method |
CN102629287A (en) * | 2012-02-29 | 2012-08-08 | 沈阳理工大学 | Automatic identification method based on standard for the exchange of product model data-compliant numerical control data interface (STEP-NC) intersection features |
Non-Patent Citations (2)
Title |
---|
张开兴 等: "面向设计重用的三维CAD 模型局部结构检索方法", 《农业机械学报》 * |
饶俊 等: "基于最大团的三维模型相似性匹配方法", 《设计与研究》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110795797A (en) * | 2019-09-26 | 2020-02-14 | 北京航空航天大学 | MBD model processing feature recognition and information extraction method |
CN110795797B (en) * | 2019-09-26 | 2021-06-18 | 北京航空航天大学 | MBD model processing feature recognition and information extraction method |
CN113806891A (en) * | 2021-09-23 | 2021-12-17 | 山东大学深圳研究院 | Rapid design method of clamp for adapting to workpiece change |
CN113806891B (en) * | 2021-09-23 | 2023-08-04 | 山东大学深圳研究院 | Quick design method of clamp suitable for workpiece change |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Estivill-Castro et al. | Amoeba: Hierarchical clustering based on spatial proximity using delaunay diagram | |
CN111639237A (en) | Electric power communication network risk assessment system based on clustering and association rule mining | |
Estivill-Castro et al. | Multi-level clustering and its visualization for exploratory spatial analysis | |
Zhang | Modeling structure and patterns in road network generalization | |
CN104573705A (en) | Clustering method for building laser scan point cloud data | |
CN110516004B (en) | Visualization method and system giving consideration to information global characteristics and local hierarchical structure | |
CN106202380A (en) | The construction method of a kind of corpus of classifying, system and there is the server of this system | |
Shi et al. | Adaptive detection of spatial point event outliers using multilevel constrained Delaunay triangulation | |
CN107944015A (en) | Threedimensional model typical structure based on simulated annealing excavates and method for evaluating similarity | |
CN115115839A (en) | Building indoor point cloud segmentation method based on local feature enhanced PointNet + + network | |
CN116775661A (en) | Big space data storage and management method based on Beidou grid technology | |
Goldin et al. | Georouting and delta-gathering: Efficient data propagation techniques for geosensor networks | |
CN116645484B (en) | Geological curved surface model construction method and device, electronic equipment and storage medium | |
Van Ham et al. | Visualization of state transition graphs | |
CN110489448A (en) | The method for digging of big data correlation rule based on Hadoop | |
Nguyen et al. | A method for efficient clustering of spatial data in network space | |
Li et al. | A multi‐scale partitioning and aggregation method for large volumes of buildings considering road networks association constraints | |
CN109241201A (en) | A kind of Laplce's centrality peak-data clustering method based on curvature | |
CN115311569A (en) | Remote sensing image-based push-fill change detection method and device and terminal equipment | |
CN104657473A (en) | Large-scale data mining method capable of guaranteeing quality monotony | |
Wang et al. | Boss recognition algorithm for application to finite element analysis | |
Chen et al. | Research and application of cluster analysis algorithm | |
Dilo et al. | Storage and manipulation of vague spatial objects using existing GIS functionality | |
Ravikumar | An Effective analysis of spatial data mining methods using range queries | |
Bae et al. | SD-Miner: A spatial data mining system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180420 |