CN103970886B - A two-stage gear cutting method similar case retrieval - Google Patents

A two-stage gear cutting method similar case retrieval Download PDF

Info

Publication number
CN103970886B
CN103970886B CN201410213426.5A CN201410213426A CN103970886B CN 103970886 B CN103970886 B CN 103970886B CN 201410213426 A CN201410213426 A CN 201410213426A CN 103970886 B CN103970886 B CN 103970886B
Authority
CN
China
Prior art keywords
process
element
example
examples
instance
Prior art date
Application number
CN201410213426.5A
Other languages
Chinese (zh)
Other versions
CN103970886A (en
Inventor
阎春平
曹卫东
肖雨亮
Original Assignee
重庆大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 重庆大学 filed Critical 重庆大学
Priority to CN201410213426.5A priority Critical patent/CN103970886B/en
Publication of CN103970886A publication Critical patent/CN103970886A/en
Application granted granted Critical
Publication of CN103970886B publication Critical patent/CN103970886B/en

Links

Abstract

本发明公开了一种两阶段滚齿工艺相似实例检索方法,其特征在于,滚齿加工工艺参数决策时,按照以下步骤进行滚齿工艺相似实例检索,具体包括步骤为:(1)实现滚齿工艺实例库的建立和存储;(2)实现检索规则和数据字典的制定;(3)实现基于表达式驱动的滚齿工艺实例一阶段检索;(4)实现基于工艺实例网络的滚齿工艺实例二阶段相似检索。 When the present invention discloses a two-stage gear cutting method similar case retrieval, characterized in that the hobbing process parameters decision, hobbing process similar case retrieving the following steps, comprises the steps of: (1) achieve hobbing establishing and storing the process instance database; (2) to achieve the development of rules and data dictionary retrieval; (3) based on a phase retrieval process example roller driving gear expression; (4) based on the implementation example hobbing process as the example of a network two-stage similar retrieval. 本发明的优点是:采用两阶段检索,利用表达式驱动初步检索得到较为合理的候选决策实例,再利用工艺实例网络进行相似检索,得到优选决策实例,提高了检索算法的检索效率和相似实例的检索准确性。 Advantage of the present invention are: two-stage retrieval, the preliminary driving by using the expression retrieved reasonable candidate decision instance, reuse similarity retrieval process example of a network, the decision result in the preferred examples, to improve the efficiency of retrieval and similarity search algorithm of an example of retrieval accuracy.

Description

一种两阶段滚齿工艺相似实例检索方法 A two-stage gear cutting method similar case retrieval

技术领域 FIELD

[0001] 本发明涉及齿轮滚齿加工技术,尤其是涉及一种滚齿加工工艺实例推理过程中对滚齿工艺相似实例检索方法。 [0001] The present invention relates to a gear hobbing technology, especially relates to a method of retrieving example hobbing process similar to the process of reasoning example hobbing process.

背景技术 Background technique

[0002] 现代滚齿加工,均是采用自动化加工机床进行加工,加工时需要根据加工要求,对加工工艺参数进行优化决策。 [0002] Modern hobbing, are automated machine tools for processing, according to the processing requirements needed for processing parameters to optimize decision-making during processing. 在滚齿加工过程中,工艺参数的智能决策的过程变得越来越重要,决策结果对齿轮加工质量和加工效率的影响受到人们广泛关注。 In the hobbing process, the process parameters of intelligent decision-making process is becoming increasingly important influence on the decision results gear machining quality and efficiency by widespread attention. 国内很多企业尚处在经验决策阶段,工艺人员依据多年加工经验,运用手册等工具,结合工件参数、工艺路线和加工要求进行工艺参数决策。 Many domestic enterprises is still in the decision-making stage experience, technical personnel based on years of experience in processing, the use of manuals and other tools, combined with the workpiece parameters, routing and processing requirements of process parameters decisions. 随着人工智能、机器学习等技术的发展,很多国内外学者将这些技术运用到工艺参数决策中,其中,具有代表性的一种技术:实例推理亦被引入其中。 With the development of artificial intelligence, machine learning technology, many scholars will use these technologies to the decision-making process parameters in which a technology representative: examples of reasoning was also introduced. 实例推理包含步骤有:检索、重用、修正实例、保存实例。 Case-Based Reasoning includes the steps of: retrieve, reuse, revise instance, saved instance. 基于实例的推理效果在某种程度上依赖于实例库的结构以及实例的知识表示方式,即在检索阶段,该阶段要达到两个目标: (1)检索出的实例尽可能少;(2)检索出来的实例尽可能与目标实例相关或相似。 Examples of inference based on the effect to some extent dependent on the structure and examples of knowledge representation database instance manner, i.e. in the retrieval phase, the phase to achieve two objectives: (1) retrieving an instance as little as possible; (2) retrieved or similar instances as relevant as possible to the target instance.

[0003] 现有实例推理在检索阶段的处理方法大多进行一阶段检索,可能出现检索出的实例与目标实例无关的情况。 [0003] Most of the conventional case-based reasoning for a method of phase retrieval in the retrieval phase of the process, where the retrieved target instance independent instance may occur. 另有一些检索方法:模板检索、分类网模型,有可能检索出多个干扰实例,给后面的修正实例、评价实例带来困难。 Some other search method: template retrieval, classification network model, it is possible to retrieve multiple instances of interference, to amend the following instances, difficult evaluation example.

发明内容 SUMMARY

[0004] 针对现有技术的不足,本发明所要解决的技术问题是,怎样提供一种能够提高滚齿工艺相似实例检索效率,提高相似实例的检索准确性的滚齿工艺相似实例检索方法,其能够实现两阶段滚齿工艺相似实例检索,实现滚齿工艺实例网络的构建,建立和存储滚齿工艺实例库,以达到为实例推理后续的工作:重用、修改实例、保存实例等提供可靠准确的优选决策实例和提高检索效率的目的。 [0004] for the deficiencies of the prior art, the present invention is the technical problem to be solved is how to provide an improved hobbing process similar case retrieval efficiency, improve the accuracy of retrieving the similar case hobbing method similar case retrieval process, which possible to realize the two-stage gear cutting similar case retrieval achieve constructed hobbing process example of a network, created and stored hobbing process instance database, in order to achieve as reasoning subsequent working examples: reuse, modified example, hold instances to provide reliable and accurate preferred examples and improve decision-making purposes retrieval efficiency.

[0005] 为了解决上述技术问题,本发明中采用了如下的技术方案: [0005] To solve the above problems, the present invention adopts the following technical solution:

[0006] -种两阶段滚齿工艺相似实例检索方法,其特征在于,基于实例推理的滚齿加工工艺参数决策时,按照以下步骤进行滚齿工艺相似实例检索,具体步骤为: [0006] - two types of gear cutting stage similar case retrieval method, wherein, when the hobbing process parameters based on CBR decision, hobbing similar case retrieval process according to the following steps, the specific steps:

[0007] (1)实现滚齿工艺实例库的建立和存储;首先,运用二元组TempCaSe(tc),tc = {KTemp,Relation}描述滚齿工艺实例库的逻辑结构,其中,KTemp是工艺实例的有限集合,K Temp ={TempIDn,ContCasen | n^O},TempIDn是KTemp的代号标示符,ContCase11 是KTemp的内容, 1^1&1:;[011是1(1'(3_上二元关系的有限集合;其次,运用物元模型描述设计工艺实例1^;^11 = (N,C,V)和问题工艺实例R_blem= (NQ,CQ,V()),其中,N表示设计实例编号,C表示设计实例的物元特征的集合,V表示特征值的集合,No表示问题实例编号,C0表示问题实例的物元特征的集合,Vo表示特征值的集合;再次,构建K Temp的内容ContCasen= {Ni,I,J,matterij,MapIsi | i 彡m, j彡k},m为设计工艺实例个数,k为物元特征个数,matterij= (Ci,vi),CieC,VieV;最后,运用图论工具建立无向加权图,即工艺实例网络来表征KTemp上二元关系的有限 [0007] (1) to achieve tooth roll created and stored in the process instance database; First, the use of tuples TempCaSe (tc), tc = {KTemp, Relation} description of the logical structure tooth roll process instance database, wherein, the process is Ktemp limited set of examples, K Temp = {TempIDn, ContCasen | n ^ O}, TempIDn is KTemp the code identifier, ContCase11 content KTemp of 1 ^ 1 & 1:; [011 1 (1 on (3_ two yuan ' finite set of relations; secondly, describe the design process example using matter element model 1 ^; ^ 11 = (N, C, V) and the process instance problems R_blem = (NQ, CQ, V ()), where, N represents a design example assemblage element characteristic number, C represents a design example, V represents the set of eigenvalues, No represents a problem instance number, C0 represents a collection of problem instance object element characteristics, Vo of represents a set of characteristic values; again, constructing K Temp of SUMMARY ContCasen = {Ni, I, J, matterij, MapIsi | i San m, j San k}, m is the number of examples of the design process, k is the number of element characteristics thereof, matterij = (Ci, vi), CieC, VieV ; Finally, the establishment of graph theory tool directed weighted graph, i.e. to characterize the process instance of the network of finite binary relation on KTemp 合此1&^011 =〈04,1>,0是工艺实例网络单元中结点,即设计实例的集合4是边的集合,¥是权值的集合; This co & ^ 1 = 011 <04,1> 0 is the process of the example network node unit, i.e. the set of design example is the set of edges 4, ¥ is a set of weights;

[0008] (2)实现检索规则和数据字典的制定;首先,将科学引文索引(SCI)检索规则与Google检索规则融合,制定的检索规则有:大小写不区分;支持布尔运算,包含与(AND)、或(0R)、非(NOT)运算,默认与(AND)运算;忽略标点符号;位置检索;截词检索;范围检索,接着,根据齿轮制造工艺手册,制定基本转化规则库,切削用量库,切削速度库,进给量库等数据字典; [0008] (2) to achieve the development of rules and retrieve data dictionary; First, the Science Citation Index (SCI) to retrieve rules integration with Google Search Rules, making retrieval rules are: case-insensitive; support Boolean operations, contains ( the aND), or (0R), non (NOT) operation, and the default (aND) operation; ignores punctuation; location search; Truncate searching; range search, then, according to the manual gear manufacturing process, development of the basic conversion rule base, cutting The amount of the library, the library cutting speed, feed rate and other data dictionary database;

[0009] (3)实现基于表达式驱动的滚齿工艺实例一阶段检索;首先,根据数据字典,针对具体滚齿工艺参数决策问题,将输入物元特征的值转化为输出物元特征的值,输出物元特征的值大部分是在某个范围内,用问题实例物元表示心^^二(NQ,CQ,V()),接着,将«转化为符合检索规则的表达式Εχρ,在滚齿工艺实例库中检索出符合表达式的设计工艺实例, 即为候选决策实例R r_it,检索出的个数为RNum; [0009] (3) implement a phase retrieval process based on the expression roller driving gear Examples; First, according to the data dictionary for the specific process parameters hobbing decision problems, the input value was converted to an output element wherein the element characteristic values ​​thereof , the output value of the element characteristics were mostly within a certain range, the problem was with the element instance represents two heart ^^ (NQ, CQ, V ()), then the «Εχρ corresponding to the search expression into rules, retrieved in gear cutting process instance database that match the design example of expressions, namely candidate decision instance R r_it, the number of retrieved RNUM;

[0010] ⑷实现基于工艺实例网络的滚齿工艺实例二阶段相似检索;首先,遍历一阶段检索得到的候选决策实例Rr_lt,利用接近度公式 [0010] ⑷ achieve a similar retrieval based on Examples hobbing process two-stage process example of a network; First, a candidate traversing the Decision Phase retrieved Rr_lt, using proximity formula

Figure CN103970886BD00061

计算结点所代表的实例第i个物元特征与问题实例心^1(«第i个物元特征之间的接近度,其次, 考虑权重的影响,计算结点Vr所代表的实例和问题实例Rprobh之间的综合接近度P(Vr,V〇), 再次,选择阈值ei (-般取0.03),根据p (vr,vo) <ei,可认为该结点vr所代表的实例为优选决策实例Ropresult,最后,利用工艺实例网络,检索含有Ropresult在工艺实例网络中代表的结点的边的权重,若小于阈值ei,取出该边的结点,得到可能优化决策实例RMpr_it。 Examples represent computing nodes i th matter element characteristics and problem instance heart ^ 1 ( «proximity between the i-th matter element wherein, secondly, considering the weight effect of the compute nodes examples and problems Vr represents examples of proximity between the integrated Rprobh P (Vr, V〇), again selecting the threshold value ei (- taken as 0.03), according to p (vr, vo) <ei, examples that may be represented by the node is preferably vr examples Ropresult decision, Finally, the process instance of the network, comprising retrieving the right side of the node in the process represented Ropresult example network heavy, if less than the threshold value ei, taken out of the edge node, may be optimized to obtain the decision RMpr_it.

[0011] 作为优化,所述步骤⑴中的工艺实例网络构建过程如下: [0011] Optimally, the process in step ⑴ example network was constructed as follows:

[0012] Stepl计数变量i = l,j = l,设计实例变量nextCase = 0; [0012] Stepl counter variable i = l, j = l, instance variables design nextCase = 0;

[0013] Step2用设计实例物元表示历史加工实例,作为无向加权图中的结点〇j; [0013] Step2 example was designed with a processing element represents the historical example, 〇j directed weighted graph of nodes as no;

[0014] Step3 当j<m,否则转到Step8; [0014] Step3 when j <m, else go Step8;

[0015] Step4 当i ,nextCase = j + 1,按公式 [0015] Step4 When i, nextCase = j + 1, according to the formula

Figure CN103970886BD00062

计算第nextCase个设计实例关于第i个物元特征与第j个设计实例第i个物元特征的接近度p (0 (nextCase) i,Oji),否则转到Step7 ; Calculation nextCase a design example on the i-th element was characterized with design examples j-th element wherein the i-th object proximity p (0 (nextCase) i, Oji), or to the Step7;

[0016] Step5按照特征权重,计算第nextCase个设计实例物元与第j个设计实例物元的综合接近度P (OnextCase,Oj),作为无向加权图中的边ek -〈OnextCase,Oj〉上的权W (nextCase) j ; [0016] Step5 According to a feature weight calculation nextCase th design examples thereof element and the j-th design examples thereof membered integrated proximity P (OnextCase, Oj), as a non-edge ek directed weighted graph of - <OnextCase, Oj> right W (nextCase) on j;

[0017] Step6 i = i+l,转到Step4; [0017] Step6 i = i + l, go Step4;

[0018] Step7 j = j+l,转到Step2; [0018] Step7 j = j + l, go to Step2;

[0019] Step8无向加权图的每个结点〇j、对应边61< =〈〇1^1;(^,〇」>、权¥(1^1;(^)」都已得到, 组成6 =〈〇3,1>,算法结束。 [0019] Step8 each node undirected weighted graph 〇j, corresponding edge 61 <= <〇1 ^ 1; (^, square ">, right ¥ (1 ^ 1; (^)" are obtained, consisting of 6 = <〇3,1> algorithm ends.

[0020] 作为优化,步骤⑴中所述的滚齿工艺实例库的逻辑结构是实例库的本质描述,是滚齿工艺相似实例检索的基础;具体构建过程如下: [0020] As optimization, a logical structure in the step ⑴ hobbing process instance database is an essential library described example, a similar process is based hobbing retrieve instances; in particular constructed as follows:

[0021] Stepl工艺实例的物元模型描述: [0021] Examples of matter element model of the process described Stepl:

[0022] 设计实例物元:设计实例是在满足特定设计要求下所获得的设计结果,其物元模型表示为: [0022] Example matter element design: Design Example is designed to result in meeting specific design requirements obtained, the object-element model represented as:

Figure CN103970886BD00071

[0024] 其中,N表示设计实例的编号;c表示设计实例的特征名;v是N关于c的量值,记作v =c (N),N,c和v称为物元R的三要素。 [0024] where, N denotes the number of design examples; wherein c denotes a design example of the name; v N value is about c, denoted by c = v (N), N, and v c was referred to as three element R elements. c和v构成的二元组M= (c,v)表示物元N的特征,称为特征元。 and c tuple M = (c, v) v represents the characteristic configuration of matter on N, called the characteristic element. 根据实例的名称N可以建立实例的索引。 The name of the instance of the N instances can be indexed.

[0025] 问题实例物元:问题实例是一组约束的集合,它反映了产品在工程语义上的设计要求,其物元模型表示为: [0025] Element problem instance was: example question is a collection of constraints, it reflects the semantics engineering product design requirements, the object-element model represented as:

Figure CN103970886BD00072

[0027] 其中,No表示问题实例的编号;CQ表示问题实例的特征名;VQ是No关于co的量值三元组,表示为VQi =〈VQil,VQih,WQi>,i = 1,2,…,η ; VQil,VQih表示特征CQi的设计需求区间;WOi表示特征CQi的权值; [0027] wherein, No represents the number of the problem instance; wherein CQ represents the name of the problem instance; No on the VQ is triple co magnitude, as represented VQi = <VQil, VQih, WQi>, i = 1,2, ..., η; VQil, VQih characterize CQi design demand interval; WOI CQi denotes the weight of the characteristic;

[0028] Step2构建工艺实例的有限集合KTemp的主要内容ContCase11: [0028] Step2 main content KTemp build a limited set of process instances ContCase11:

[0029] 将设计工艺实例用物元描述后,得到每个设计工艺实例的特征元,作为ContCase11 的主体,从而将ContCase11表示为{Ni,I,J,matterij,MapIsi | i<m, j<k},其中,m为设计工艺实例个数,k为物元特征个数,依据编号N,依次将设计工艺实例放入ContCase11中,matteriN -(ci,Vi) ,CiGCjViGV; [0029] Examples of the design process is described matter element, wherein each process instance to obtain design element, as a host ContCase11, thereby ContCase11 represented as {Ni, I, J, matterij, MapIsi | i <m, j < k}, where, m is the number of instances the design process, k is the element number of the characteristics thereof, according to the number N, the order of the design process instances into ContCase11, matteriN - (ci, Vi), CiGCjViGV;

[0030] Step3运用权力要求2中建立的工艺实例网络来表征KTempl二元关系的有限集合Relation =〈0,E,W>,0是工艺实例网络单元中结点(设计实例)的集合,E是边的集合,W是权值的集合; [0030] Step3 process requires the use of power in Example 2 to establish the network to characterize the finite set of binary relations Relation KTempl = <0, E, W>, 0 is set in the process of the example network node element (design example), E is the set of edges, W is a set of weights;

[0031] Step4运用二元组TempCase (tc),tc= {KTemp,Relation}描述滚齿工艺实例库的逻辑结构,其中,KTemp= {TempIDn,ContCasen|n彡0},TempIDn是K Te19代号标示符,ContCase11 ={Ni,I ,J,matterij,MapIsi | i^im, j^ik} ,Relation = <0,E,ff>〇 [0031] Step4 use tuple TempCase (tc), tc = {KTemp, Relation} description of the logical structure roller gear case library technology, where, KTemp = {TempIDn, ContCasen | n San 0}, TempIDn code is marked K Te19 Fu, ContCase11 = {Ni, I, J, matterij, MapIsi | i ^ im, j ^ ik}, Relation = <0, E, ff> square

[0032] 作为优化,所述步骤(3)中的基于表达式驱动的滚齿工艺实例一阶段检索的具体步骤包括: [0032] Optimally, the step (3) the specific example of process step roller gear stage based on the retrieved driving expression comprises:

[0033] Stepl根据数据字典,针对具体滚齿工艺参数决策问题,将输入物元特征包括工件类别、法向模数、齿数、压力角、螺旋角、材料、精度、径向变位系数、滚切方式的值转化为输出物元特征包括滚刀类别、精度、头数、螺旋升角、滚刀转速、轴向进给速度、径向进给速度、 滚切余量、进给量、切削液的值,输出物元特征的值大部分是在某个范围内,用问题实例物兀1 表小Rproblem- (N〇, C〇, V〇); [0033] Stepl The data dictionary, hobbing process parameters for specific decision problems, the feature of the input element includes a workpiece matter category, the normal module, number of teeth, pressure angle and helix angle, the material, the accuracy, the radial displacement coefficient, roller value cut system into an output element thereof characterized in comprising a hob type, precision, number of heads, helix angle, hob rotational speed, axial feed velocity, the radial feed speed, rolling balance cut, feed rate, cutting value elements characteristic value of the liquid output was mostly within a certain range, the problem with the example in table 1 was small Wu Rproblem- (N〇, C〇, V〇);

[0034] Step2根据检索规则,制定检索表达式,具体表示为:Exp = { (cqi =滚齿类别,νοι = (圆柱斜齿轮,圆柱斜齿轮))AND (cQ2 =滚齿精度,VQ2 = (AA,AAA)) AND (cQ3 =滚齿头数,v03 = (1,2))AND…; [0034] Step2 according to a search rule to develop search expression, specifically expressed as: Exp = {(cqi = hobbing category, νοι = (cylindrical helical gears, cylindrical helical gears)) AND (cQ2 = hobbing precision, VQ2 = ( AA, AAA)) AND (cQ3 = hobbing head number, v03 = (1,2)) AND ...;

[0035] Step3检索滚齿工艺实例库中符合表达式Exp的设计工艺实例,即为候选决策实例Rr_it,假定检索出的实例个数为RNum; [0035] Step3 retrieval gear cutting process instance database that match the expression Exp example design, namely the Decision candidate Rr_it, assume that the number of examples is retrieved RNUM;

[0036] 作为优化,所述步骤(4)中的基于工艺实例网络的滚齿工艺实例二阶段相似检索的具体步骤包括: [0036] Optimally, the step (4) based on the specific example of process steps hobbing two-stage search process similar example of a network comprising:

[0037] stepl计数变量i = 1,m= 1; [0037] stepl counting variable i = 1, m = 1;

[0038] Step2当r^iRNum,否则转到Stepll; [0038] Step2 When r ^ iRNum, else go Stepll;

[0039] Step3当i<n (η为问题物元特征个数),否则转到Step6; [0039] Step3 When i <n (η is a problem wherein the number of element thereof), or go to Step6;

[0040] Step4,用公式 [0040] Step4, using the formula

Figure CN103970886BD00081

计算结点Vi•所代表的实例第i个物元特征与问题实例第i个物元特征之间的接近度P (Vri,V0i); Wherein the i-th element instance was calculated nodes Vi • represents the problem instance proximity between the i-th element thereof wherein P (Vri, V0i);

[0041] Step5 i = i+l,转到Step3; [0041] Step5 i = i + l, go to Step3;

[0042] S tep 6按照特征权重,计算结点vr所代表的实例物元和问题实例物元之间的综合接近度P (Vr,V0); [0042] S tep 6 characterized in accordance with weights of the compute nodes and the element instance was the problem instance was represented vr integrated proximity between the element P (Vr, V0);

[0043] Step7选定阈值ei,若p (Vr,VQ) <ei,可认为该结点为优选决策实例,转到Step8,否则转到Step2; [0043] Step7 selected threshold ei, if p (Vr, VQ) <ei, the node may be considered as examples of preferred decision, to Step8, otherwise go to Step2;

[0044] Step8若m〈 = NodeNum (NodeNum为结点个数)而且m辛r,计算元优选决策实例结点相连接的结点权值Wrm (m的值在结点总个数范围内,m#r)与P (Vr, VQ)相减后的绝对值I Wrm-P (vr,vo) I <ei,则提取该结点为可能优选决策实例,否则,转到Stepll。 [0044] Step8 if m <= NodeNum (NodeNum is the number of nodes) and oct m r, node weight calculation element is preferably coupled to node examples decision value within the range of the total number Wrm * (the value of m in the node, m # r) and the absolute value of P (Vr, VQ) subtraction I Wrm-P (vr, vo) I <ei, is extracted as the node may be preferred example decision, otherwise, go to Stepll.

[0045] Step9 m=m+l,转到Step8; [0045] Step9 m = m + l, go Step8;

[0046] SteplO r = r+l,转到Step2; [0046] SteplO r = r + l, go to Step2;

[0047] Stepll得到优选决策实例和可能优选决策实例就是检索出来的结果,算法结束。 [0047] Stepll obtained Optimum examples and examples that may be preferably retrieved decision result, the algorithm ends.

[0048] 本发明中采用两阶段检索,利用表达式驱动初步检索得到较为合理的候选决策实例,再利用工艺实例网络进行相似检索,提高了检索算法的检索效率和相似实例的检索准确性。 [0048] In the present invention, two-stage retrieval, the preliminary driving by using the expression retrieved reasonable candidate decision instance, reuse similarity retrieval process example of a network to improve the efficiency of the search algorithm and the retrieval of the similar case retrieval accuracy. 本发明的一种两阶段滚齿工艺相似实例检索方法,基于表达式驱动和工艺实例网络等技术而实现,该滚齿加工工艺相似实例检索中,抓住滚齿加工工艺实例的特点,运用物元模型描述工艺实例,利用图论工具构建工艺实例网络,将两者结合,组建二元组TempCase (tc),tc= {KTemp,Relation},即为滚齿工艺实例库的逻辑结构,用该结构将工艺实例存储到数据库,运用表达式驱动实现一阶段检索,运用工艺实例网络实现二阶段检索得到优选决策实例;根据优选决策实例进行实例推理时,即可达到提高检索准确性和提高检索效率的效果。 One inventive two-stage gear cutting method similar case retrieval, based on the expression and the driving example of process technology networks is achieved, the hobbing process similar case retrieval, seize hobbing process instance processing characteristics, the use thereof metamodel described process example, the process graph theory tool to build a network example, the combination of the two, the formation tuple TempCase (tc), tc = {KTemp, Relation}, that is, the logical structure of the tooth roll process instance database with which the structure of the process instances stored in the database, the use of the expression phase of the drive to achieve a search, using a two-stage process instance networks retrieved Optimum example; when the decision preferred examples of reasoning, and to improve the accuracy of retrieval to improve retrieval efficiency Effect.

附图说明 BRIEF DESCRIPTION

[0049] 图1为本发明具体实施方式中两阶段滚齿工艺相似实例检索过程的示意图; [0049] FIG 1 two-stage embodiment hobbing process diagram showing an example similar to the retrieval process of the present invention;

[0050] 图2为本发明具体实施方式中工艺实例网络的示意图; [0050] FIG. 2 is a schematic process example of a network specific embodiment the invention;

[0051] 图3为本发明具体实施方式中工艺实例网络构建过程的示意图; [0051] FIG. 3 example embodiment network diagram during the process of the present invention is constructed;

[0052]图4为本发明具体实施方式中基于表达式驱动的滚齿工艺实例一阶段检索过程的示意图; [0052] Fig 4 a schematic view of a particular embodiment based retrieval process example hobbing process stage of the present invention, expression is driven;

[0053] 图5为本发明具体实施方式中基于工艺实例网络的滚齿工艺实例二阶段相似检索过程的示意图。 [0053] FIG. 5 is a schematic process based similarity retrieval process example hobbing process example of a network of two-stage embodiment of the present invention.

具体实施方式 Detailed ways

[0054] 本发明的思路是:抓住滚齿加工工艺实例的特点,运用物元模型描述工艺实例,利用图论工具构建工艺实例网络,将两者结合,组建二元组TempCa Se(tc),tc= {KTemp, Relation},即为滚齿工艺实例库的逻辑结构,用该结构将工艺实例存储到数据库,运用表达式驱动实现一阶段检索,运用工艺实例网络实现二阶段检索得到优选决策实例。 [0054] The idea of ​​the invention is: to seize the hobbing process characteristics of process instances, the use of the process described in example element model was constructed by using the example network graph theory process tool, a combination of both, the formation tuple TempCa Se (tc) , tc = {KTemp, Relation}, that is, the logical structure of the tooth roll process instance database, with this process instance will be stored in the configuration database, the use of the expression phase of the drive to achieve a search, using a two-stage process instance networks retrieved Optimum instance.

[0055] 下面结合附图对本发明作进一步说明: [0055] DRAWINGS The present invention is further described:

[0056] 本发明为一种两阶段滚齿工艺相似实例检索方法,本方法中基于实例推理的滚齿加工工艺参数决策时,按照此方法进行相似实例检索,如图1-图5所示,包括以下具体步骤: [0057] (1)实现滚齿工艺实例库的建立和存储;首先,运用二元组TempCa Se(tc),tc = {KTemp,Relation}描述滚齿工艺实例库的逻辑结构,其中,KTemp是工艺实例的有限集合,K Temp ={TempIDn,ContCase111 η彡0},TempIDn是KTe19代号标示符,ContCase11 是KTe19 内容, Relation是KTemp上二元关系的有限集合;其次,运用物元模型描述设计工艺实例Rdeslgn = (N,C,V)和问题工艺实例R_biem= (NQ,CQ,V()),其中,N表示设计实例编号,C表示设计实例的物元特征的集合,V表示特征值的集合,No表示问题实例编号,Co表示问题实例的物元特征的集合,Vo表示特征值的集合;再次,构建K Temp的内容ContCasen= {Ni,I,J,matterij,Ma [0056] The present invention is a two stage gear cutting method similar case retrieval, hobbing process parameters during the Decision based reasoning, a similar case retrieval method according to the present method, as shown in Figures 1 to 5, It comprises the following steps: [0057] (1) to achieve tooth roll created and stored in the process instance database; first, the use of tuples TempCa Se (tc), tc = {KTemp, Relation} tooth process described logical structure of the roller case library wherein, Ktemp is a finite set of process instances, K Temp = {TempIDn, ContCase111 η San 0}, TempIDn is KTe19 code identifier, ContCase11 is KTe19 content, the relation is a finite set of binary relations on Ktemp; secondly, the use thereof described the design process example metamodel Rdeslgn = (N, C, V) and the process instance problems R_biem = (NQ, CQ, V ()), where, N represents a design example number, C represents assemblage element feature and examples, V represents the set of eigenvalues, No represents a problem instance number, Co represents assemblage element characteristic problem instance, Vo of represents a set of characteristic values; re-construct the contents of K Temp of ContCasen = {Ni, I, J, matterij, Ma pIsi | i 彡m, j彡k},m为设计工艺实例个数,k为物元特征个数,matterij= (ci,vi),cieC,vieV;最后,运用图论工具建立工艺实例网络(无向加权图)来表征以^上二元关系的有限集合Relation =〈0,E,W>,0是工艺实例网络单元中结点(设计实例)的集合,E是边的集合,W是权值的集合; pIsi | i San m, j San k}, m is the number of examples of the design process, k is the number of element characteristics thereof, matterij = (ci, vi), cieC, vieV; finally, the use of tools to create process instances graph theory networks ( no characterized in the relation to the directed weighted graph ^) a limited set of relation = <0, E, W>, 0 is set in the process of the example network node element (design example), E is the set of edges, W is a set of weights;

[0058] (2)实现检索规则和数据字典的制定;首先,将科学引文索引(SCI)检索规则与Google检索规则融合,制定的检索规则有:大小写不区分;支持布尔运算,包含与(AND)、或(0R)、非(NOT)运算,默认与(AND)运算;忽略标点符号;位置检索;截词检索;范围检索,接着,根据齿轮制造工艺手册,制定基本转化规则库,切削用量库,切削速度库,进给量库等数据字典; [0058] (2) to achieve the development of rules and retrieve data dictionary; First, the Science Citation Index (SCI) to retrieve rules integration with Google Search Rules, making retrieval rules are: case-insensitive; support Boolean operations, contains ( the aND), or (0R), non (NOT) operation, and the default (aND) operation; ignores punctuation; location search; Truncate searching; range search, then, according to the manual gear manufacturing process, development of the basic conversion rule base, cutting The amount of the library, the library cutting speed, feed rate and other data dictionary database;

[0059] (3)实现基于表达式驱动的滚齿工艺实例一阶段检索;首先,根据数据字典,针对具体滚齿工艺参数决策问题,将输入物元特征的值转化为输出物元特征的值,输出物元特征的值大部分是在某个范围内,用问题实例物元表示心^^二(NQ,CQ,V()),接着,将«转化为符合检索规则的表达式Εχρ,在滚齿工艺实例库中检索出符合表达式的设计工艺实例, 即为候选决策实例R r_it,检索出的个数为RNum; [0059] (3) implement a phase retrieval process based on the expression roller driving gear Examples; First, according to the data dictionary for the specific process parameters hobbing decision problems, the input value was converted to an output element wherein the element characteristic values ​​thereof , the output value of the element characteristics were mostly within a certain range, the problem was with the element instance represents two heart ^^ (NQ, CQ, V ()), then the «Εχρ corresponding to the search expression into rules, retrieved in gear cutting process instance database that match the design example of expressions, namely candidate decision instance R r_it, the number of retrieved RNUM;

[0060] ⑷实现基于工艺实例网络的滚齿工艺实例二阶段相似检索;首先,遍历一阶段检索得到的候选决策实例Rr_lt,利用接近度公式 [0060] ⑷ achieve a similar retrieval based on Examples hobbing process two-stage process example of a network; First, a candidate traversing the Decision Phase retrieved Rr_lt, using proximity formula

Figure CN103970886BD00091

十算结点所代表的实例第i个物元特征与问题实例心^1(«第i个物元特征之间的接近度,其次, 考虑权重的影响,计算结点Vr所代表的实例和问题实例Rprobh之间的综合接近度P(Vr,V〇), 再次,选择阈值ei (-般取0.03),根据P (Vr,V0) < ei,可认为该结点Vr所代表的实例为优选决策实例Ropresult,最后,利用工艺实例网络,检索含有Ropresult在工艺实例网络中代表的结点的边的权重,若小于阈值ei,取出该边的结点,得到可能优化决策实例RMpr_it; Example 1 (proximity between the i-th object element wherein «, secondly, considering weight impact examples represent ten Operators node i th matter element characteristics and problem instance heart ^, calculated nodes Vr represents and integrated proximity between the problem instance Rprobh P (Vr, V〇), again selecting the threshold value ei (- taken as 0.03), according to P (Vr, V0) <ei, examples that may be represented by the node is Vr Optimum examples Ropresult, Finally, the process instance of the network, retrieval comprising the right side of the node Ropresult represented in the process example network heavy, if less than the threshold value ei, remove the node of the edge, to obtain possible to optimize the decision RMpr_it;

[0061] 本发明是基于表达式驱动和工艺实例网络,来实现滚齿工艺相似实例检索的,具体过程如图1所示。 [0061] The present invention is based on the expression and processing instances drive network, to achieve a similar case retrieval hobbing process, the specific procedure shown in Fig.

[0062] 上述步骤1中,ContCase11的具体结构如表1所示: [0062] In the above step 1, ContCase11 specific structures as shown in Table 1:

[0063] 表1 ContCase11的具体结构 [0063] Table 1 ContCase11 specific configuration of the

Figure CN103970886BD00101

[0065] 定义1工艺实例网络是一个无向加权图G =〈V,E,W>,其中V是工艺实例网络中结点的集合,结点是用物元模型描述的实例物元。 [0065] The process defines an example of a network is an undirected weighted graph G = <V, E, W>, where V is the set of process instances in a network node, the node is an example of matter element model element was described. E是边的集合,W是权值的集合表示实例物元丨,边61<=〈¥1,¥」>££表示任务;[与」之间的联系。 E is the set of edges, W is a set of weights shows an example of matter element Shu, edge 61 <= <¥ 1, ¥ "> ££ represents the task; link between [and." 边61 {上的权¥1表示边上的权值(>1,¥」之间的接近度)。 On the right side 61 is ¥ {weight (> 1, the proximity between ¥ ") represents an edge. 工艺实例网络示意图如图2所示。 Examples of network schematic process shown in Figure 2.

[0066] 上述步骤1中,工艺实例网络(无向加权图)的构建过程如图3所示,具体步骤如下: [0066] Step 1 above, the process instance of the network (FIG weighted undirected) construction process shown in Figure 3, the following steps:

[0067] Stepl计数变量i = l,j = l,设计实例变量nextCase = 0; [0067] Stepl counter variable i = l, j = l, instance variables design nextCase = 0;

[0068] Step2用设计实例物元表示历史加工实例,作为无向加权图中的结点〇j; [0068] Step2 example was designed with a processing element represents the historical example, 〇j directed weighted graph of nodes as no;

[0069] Step3 当j<m,否则转到Step8; [0069] Step3 when j <m, else go Step8;

[0070] Step4当i ,nextCase = j + 1,按公式 [0070] Step4 When i, nextCase = j + 1, according to the formula

Figure CN103970886BD00102

,计算第nextCase个设计实例关于第i个物元特征与第j个设计实例第i个物元特征的接近度p (〇(nextcase)i,〇ji),否贝1J转到Step7; Calculating a design example of nextCase on the i-th element was characterized with design examples j-th element wherein the i-th object proximity p (square (nextcase) i, 〇ji), whether the shellfish to the Step7 1J;

[0071] Step5按照特征权重,计算第nextCase个设计实例物元与第j个设计实例物元的综合接近度P (OnextCase,Oj),作为无向加权图中的边ek =〈OnextCase,〇j>上的权w (nextCase) j; [0071] Step5 According to a feature weight calculation nextCase th design examples thereof element and the j-th design examples thereof membered integrated proximity P (OnextCase, Oj), as a non-directional edge ek weighting figures = <OnextCase, 〇j > weight w (nextCase) on J;

[0072] Step6 i = i+l,转到Step4; [0072] Step6 i = i + l, go Step4;

[0073] Step7 j = j+l,转到Step2; [0073] Step7 j = j + l, go to Step2;

[0074] Step8无向加权图的每个结点〇j、对应边61< =〈〇1^1;(^,〇』>、权¥(1^1;(^)」都已得到, 组成6 =〈〇3,1>,算法结束。 [0074] Step8 each node undirected weighted graph 〇j, corresponding edge 61 <= <〇1 ^ 1; (^, square ">, right ¥ (1 ^ 1; (^)" are obtained, consisting of 6 = <〇3,1> algorithm ends.

[0075] 上述步骤1中,滚齿工艺实例库的建立过程如图3所示,具体步骤如下: [0075] Step 1 above, hobbing process instance database setup procedure shown in Figure 3, the following steps:

[0076] Stepl工艺实例的物元模型描述: [0076] Examples of matter element model of the process described Stepl:

[0077] 设计实例物元:设计实例是在满足特定设计要求下所获得的设计结果,其物元模型表示为: [0077] Example matter element design: Design Example is designed to result in meeting specific design requirements obtained, the object-element model represented as:

Figure CN103970886BD00111

[0079] 其中,N表示设计实例的编号;c表示设计实例的特征名;v是N关于c的量值,记作v =c (N),N,c和v称为物元R的三要素。 [0079] where, N denotes the number of design examples; wherein c denotes a design example of the name; v N value is about c, denoted by c = v (N), N, and v c was referred to as three element R elements. c和v构成的二元组M= (c,v)表示物元N的特征,称为特征元。 and c tuple M = (c, v) v represents the characteristic configuration of matter on N, called the characteristic element. 根据实例的名称N可以建立实例的索引。 The name of the instance of the N instances can be indexed.

[0080] 问题实例物元:问题实例是一组约束的集合,它反映了产品在工程语义上的设计要求,其物元模型表示为: [0080] Element problem instance was: example question is a collection of constraints, it reflects the semantics engineering product design requirements, the object-element model represented as:

Figure CN103970886BD00112

[0082] 其中,No表示问题实例的编号;CQ表示问题实例的特征名;VQ是No关于CQ的量值三元组,表示为VQi =〈VQil,VQih,WQi>,i = 1,2,…,η ; VQil,VQih表示特征CQi的设计需求区间;WOi表示特征CQi的权值; [0082] wherein, No represents the number of the problem instance; wherein CQ represents the name of the problem instance; No is the VQ on the magnitude of triplet CQ, expressed as VQi = <VQil, VQih, WQi>, i = 1,2, ..., η; VQil, VQih characterize CQi design demand interval; WOI CQi denotes the weight of the characteristic;

[0083] Step2构建工艺实例的有限集合KTemp的主要内容ContCase11: [0083] Step2 main content KTemp build a limited set of process instances ContCase11:

[0084] 将设计工艺实例用物元描述后,得到每个设计工艺实例的特征元,作为ContCase11 的主体,从而将ContCase11表示为{Ni,I,J,matterij,MapIsi | i<m, j<k},其中,m为设计工艺实例个数,k为物元特征个数,依据编号N,依次将设计工艺实例放入ContCase11中,matteriN -(ci,Vi) ,CiGCjViGV; [0084] Examples of the design process is described matter element, wherein each process instance to obtain design element, as a host ContCase11, thereby ContCase11 represented as {Ni, I, J, matterij, MapIsi | i <m, j < k}, where, m is the number of instances the design process, k is the element number of the characteristics thereof, according to the number N, the order of the design process instances into ContCase11, matteriN - (ci, Vi), CiGCjViGV;

[0085] Step3运用权力要求2中建立的工艺实例网络来表征KTempl二元关系的有限集合Relation =〈0,E,W>,0是工艺实例网络单元中结点(设计实例)的集合,E是边的集合,W是权值的集合; [0085] Step3 process requires the use of power in Example 2 to establish the network to characterize the finite set of binary relations Relation KTempl = <0, E, W>, 0 is set in the process of the example network node element (design example), E is the set of edges, W is a set of weights;

[0086] Step4运用二元组TempCase (tc),tc= {KTemp,Relation}描述滚齿工艺实例库的逻辑结构,其中,KTemp= {TempIDn,ContCasen|n彡0},TempIDn是K Te19代号标示符,ContCase11 ={Ni,I ,J,matterij,MapIsi | i^im, j^ik} ,Relation = <0,E,ff>〇 [0086] Step4 use tuple TempCase (tc), tc = {KTemp, Relation} description of the logical structure roller gear case library technology, where, KTemp = {TempIDn, ContCasen | n San 0}, TempIDn code is marked K Te19 Fu, ContCase11 = {Ni, I, J, matterij, MapIsi | i ^ im, j ^ ik}, Relation = <0, E, ff> square

[0087] 上述步骤2中,检索规则和数据字典如表2-表10所示。 [0087] Step 2 above, to retrieve rules and data dictionary as shown in Table 2 - Table 10.

[0088] 表2检索规则 [0088] Table 2 Search Rules

Figure CN103970886BD00121

[0090] 表3输入输出基本转化规则库 [0090] Table 3 basic input-output conversion rule base

Figure CN103970886BD00122

[0092] 表4普通高速钢滚刀切削用量 [0092] TABLE 4 Normal speed steel hob cutting parameters

[0093] [0093]

Figure CN103970886BD00131

[0094] 表5滚切不同材质齿轮时切削速度 Cutting speed [0094] Table 5 different materials gear hobbing

Figure CN103970886BD00132

[0096] 表6滚切不同材质齿轮时的进给量 Feed amount of [0096] TABLE 6 different materials gear hobbing

Figure CN103970886BD00133

[0099] 表7滚切余量分配表 [0099] Table 7 hobbing margin allocation table

Figure CN103970886BD00134

[0101] 表8硬齿面齿轮的刮削留量 [0101] Table 8 Hardened gear shaving allowance

Figure CN103970886BD00135

[0103] 表9滚刀材料选择规则 [0103] TABLE 9 Material selection rule hob

Figure CN103970886BD00136

[0105] 表10滚刀精度选择规则 [0105] Table 10 hob accuracy selection rule

Figure CN103970886BD00137

[0107] 上述步骤3中,基于表达式驱动的滚齿工艺实例一阶段检索的过程如图4所示,具体步骤如下: [0107] In the above step 3, a phase retrieval process based on the expression of the drive roller teeth process Example 4, the following steps:

[0108] Stepl根据数据字典,针对具体滚齿工艺参数决策问题,将输入物元特征(工件类另IJ、法向模数、齿数、压力角、螺旋角、材料、精度、径向变位系数、滚切方式)的值转化为输出物元特征(滚刀类别、精度、头数、螺旋升角、滚刀转速、轴向进给速度、径向进给速度、滚切余量、进给量、切削液)的值,输出物元特征的值大部分是在某个范围内,用问题实例物元表/jNRproblem- (N〇, C〇, V〇); [0108] Stepl The data dictionary, hobbing process parameters for specific decision problems, wherein the input element composition (IJ workpiece another class, the normal module, number of teeth, pressure angle and helix angle, the material, the accuracy, the radial displacement coefficient , hobbing method) values ​​were converted to element characteristics (hob type, precision, output product number of heads, helix angle, hob rotational speed, axial feed velocity, the radial feed speed, hobbing balance, feed most characteristic value amount value elements, the cutting fluid), the output thereof is within a certain range, the problem was with the examples in table element / jNRproblem- (N〇, C〇, V〇);

[0109] 3丨6?2根据检索规则,制定检索表达式,具体表示为41卩={(〇()1 =滚齿类别,¥〇1 = (圆柱斜齿轮,圆柱斜齿轮))AND (CQ2 =滚齿精度,VQ2 = (AA,AAA)) AND (CQ3 =滚齿头数,V03 = (1,2))AND…; [0109] Shu 3 6? 2 according to a search rule to develop search expression, particularly denote Jie 41 = {(square () = 1 hobbing category, ¥ = 〇1 (cylindrical helical gears, cylindrical helical gears)) the AND ( CQ2 = hobbing precision, VQ2 = (AA, AAA)) AND (CQ3 = hobbing head number, V03 = (1,2)) AND ...;

[0110] Step3检索滚齿工艺实例库中符合表达式Exp的设计工艺实例,即为候选决策实例Rr_it,假定检索出的实例个数为RNum; [0110] Step3 retrieval gear cutting process instance database that match the expression Exp example design, namely the Decision candidate Rr_it, assume that the number of examples is retrieved RNUM;

[0111] 上述步骤4中,基于工艺实例网络的滚齿工艺实例二阶段相似检索的过程如图5所示,具体步骤如下: [0111] In the above step 4, a two-stage process of hobbing process instance based similarity retrieval process example of a network shown in Figure 5, the following steps:

[0112] stepl计数变量i = 1,m= 1,r = 1; [0112] stepl counting variable i = 1, m = 1, r = 1;

[0113] 3七6卩2当1'<1^1皿,否则转到3七6卩11; [0113] Jie 2 3 6 When seven 1 '<1 1 ^ dish, seven 3 or 6 to 11 Jie;

[0114] Step3 当i<k,否则转到Step6; [0114] Step3 When i <k, otherwise go to Step6;

[0115] Step4,用公式 [0115] Step4, using the formula

Figure CN103970886BD00141

计算结点所代表的实例第i个物元特征与问题实例第i个物元特征之间的接近度P (Vri,V0i); Wherein the i-th element composition represented by the calculated examples of the problem instance node proximity between the i-th element thereof wherein P (Vri, V0i);

[0116] Step5 i = i+l,转到Step3; [0116] Step5 i = i + l, go to Step3;

[0117] Step6按照表11给出的权重,计算结点vr所代表的实例物元和问题实例物元之间的综合接近度P(Vr,V0); [0117] Step6 right weight given in Table 11, Examples of the compute nodes and the element composition represented by vr problems Examples was integrated proximity between the element P (Vr, V0);

[0118] Step7选定阈值ei,若p (vr,vq) <ei,可认为该结点为优选决策实例,转到Step8,否则转到SteplO [0118] Step7 selected threshold ei, if p (vr, vq) <ei, the node may be considered as examples of preferred decision, to Step8, else go SteplO

[0119] Step8若m〈 = NodeNum (NodeNum为结点个数)而且m辛r,计算元优选决策实例结点相连接的结点权值Wrm (m的值在结点总个数范围内,m#r)与P (Vr, VQ)相减后的绝对值I Wrm-P (vr,vo) I <ei,则提取该结点为可能优选决策实例,否则,转到Stepll。 [0119] Step8 if m <= NodeNum (NodeNum is the number of nodes) and oct m r, node weight calculation element is preferably coupled to node examples decision value within the range of the total number Wrm * (the value of m in the node, m # r) and the absolute value of P (Vr, VQ) subtraction I Wrm-P (vr, vo) I <ei, is extracted as the node may be preferred example decision, otherwise, go to Stepll.

[0120] Step9 m=m+l,转到Step8; [0120] Step9 m = m + l, go Step8;

[0121] SteplO r = r+l,转到Step2; [0121] SteplO r = r + l, go to Step2;

[0122] Stepll得到优选决策实例和可能优选决策实例就是检索出来的结果,算法结束。 [0122] Stepll obtained Optimum examples and examples that may be preferably retrieved decision result, the algorithm ends.

[0123] 表11实例物元输出特征权重 [0123] Table 11 Example Element outputs feature weights was

Figure CN103970886BD00142

Claims (5)

1. 一种两阶段滚齿工艺相似实例检索方法,其特征在于,基于实例推理的滚齿加工工艺参数决策时,按照以下步骤进行滚齿工艺相似实例检索,具体步骤为: ⑴实现滚齿工艺实例库的建立和存储;首先,运用二元组TempCaSe(tc),tc= {KTemp, Relation}描述滚齿工艺实例库的逻辑结构,其中,KTemp是工艺实例的有限集合,KTemp = {TempIDn,ContCase111 η彡0},TempIDn是KTemi^代号标示符,ContCase11 是KTemi^ 内容, Relation是KTemp上二元关系的有限集合;其次,运用物元模型描述设计工艺实例Rdeslgn = (N,C,V)和问题工艺实例R_biem= (NQ,CQ,V()),其中,N表示设计实例编号,C表示设计实例的物元特征的集合,V表示特征值的集合,No表示问题实例编号,Co表示问题实例的物元特征的集合,Vo表示特征值的集合;再次,构建K Temp的内容ContCasen= {Ni,I,J,matterij,MapIsi | i 彡m, j彡k},m为设 A two-stage gear cutting method similar case retrieval, characterized in that the hobbing process based on CBR decision parameters, hobbing similar case retrieval process according to the following steps, specifically steps: ⑴ achieve hobbing process examples of libraries created and stored; first, the use of tuples TempCaSe (tc), tc = {KTemp, Relation} description of the logical structure roller gear case library technology wherein, Ktemp is a finite set of process instances, KTemp = {TempIDn, San ContCase111 η 0}, TempIDn KTemi ^ is the code identifier, ContCase11 KTemi ^ is the content, the relation is a finite set of binary relations on Ktemp; secondly, the use of the design process was described in example metamodel Rdeslgn = (N, C, V) and example of process issues R_biem = (NQ, CQ, V ()), where, N represents a design example number, C represents assemblage element characteristic design example, V represents the set of eigenvalues, No represents a problem instance number, Co represents element problems characteristic assemblage example, represents a set Vo of eigenvalues; re-construct the contents of K Temp ContCasen = {Ni, I, J, matterij, MapIsi | i San m, j San k}, m is set 计工艺实例个数,k为物元特征个数,matterij= (CNi,j,VNi,j),CNi,jeC, VN^eV;最后,运用图论工具建立无向加权图,即工艺实例网络来表征KTempl二元关系的有限集合1^1 &^〇11 =〈04,1>,0是工艺实例网络单元中结点,即设计实例的集合4是边的集合,W是权值的集合; (2) 实现检索规则和数据字典的制定;首先,将科学引文索引(SCI)检索规则与Google 检索规则融合,制定的检索规则有:大小写不区分;支持布尔运算,包含与(AND)、或(OR)、非(NOT)运算,默认与(AND)运算;忽略标点符号;位置检索;截词检索;范围检索,接着,根据齿轮制造工艺手册,制定输入输出基本转化规则库,切削用量库,切削速度库,进给量库数据字典; (3) 实现基于表达式驱动的滚齿工艺实例一阶段检索;首先,根据数据字典,针对具体滚齿工艺参数决策问题,将输入物元特征的值转化为输出物元 Count the number of process instances, k is the element number of the characteristics thereof, matterij = (CNi, j, VNi, j), CNi, jeC, VN ^ eV; Finally, the establishment of graph theory tool directed weighted graph, i.e., the process instance of the network characterized finite set of binary relations KTempl 1 ^ 1 & ^ 〇11 = <04,1> 0 is the process of the example network node unit, i.e. the set of design example is the set of edges 4, W is a set of weights ; (2) to achieve the development of rules and retrieve data dictionary; first, the Science Citation index (SCI) to retrieve rules integration with Google Search rules, making retrieval rules are: case-insensitive; support Boolean operations, contains (aND) , or (oR), non (NOT) operation, and the default (aND) operation; ignores punctuation; location search; Truncate searching; range search, then, according to the manual gear manufacturing process, development of the basic input-output conversion rule base, cutting the amount of the library, the library cutting speed, feed amount data dictionary database; (3) implement a hobbing process instance based on the phase retrieval expression driven; first, according to the data dictionary for the specific process parameters hobbing decision problems, the input element was characteristic value into an output element thereof 征的值,输出物元特征的值大部分是在某个范围内,用问题实例物元表示(NQ,CQ,V()),接着,将Rproblem转化为符合检索规则的表达式Exp,在滚齿工艺实例库中检索出符合表达式的设计工艺实例,即为候选决策实例R r_it,检索出的个数为RNum; ⑷实现基于工艺实例网络的滚齿工艺实例二阶段相似检索;首先,遍历一阶段检索得到的候选决策实例Rresult,利用接近度 Intrinsic value, the output value of the element characteristics were mostly within a certain range, the problem with the example was represented by element (NQ, CQ, V ()), then into the Rproblem corresponding to the search expressions Exp rules, in hobbing process instance database retrieving process meet the design example of expressions, namely candidate decision instance R r_it, the number of retrieved RNum; ⑷ achieve a similar retrieval based on the hobbing process example of a process example of a network of the second stage; first, examples of candidate decision stage traverse retrieved Rresult, using proximity
Figure CN103970886BC00021
Vr所代表的实例第i个物元特征与问题实例Rproblm第i个物元特征之间的接近度,其次,考虑权重的影响,计算结点Vr所代表的实例和问题实例Rproblem之间的综合接近度P (Vr,VQ),再次,选择阈值ei,根据P (vr,vo) <ei,可认为该结点vj/f代表的实例为优选决策实例Result, 最后,利用工艺实例网络,检索含有Rc^esult在工艺实例网络中代表的结点的边的权重,若小于阈值ei,取出该边的结点,得到可能优化决策实例RMprasult。 Proximity between the i-th element matter was characterized with i-th element Rproblm problem instance wherein Vr represents examples, secondly, to consider the influence of weight, calculated nodes Vr represents examples and synthesis examples of problems between Rproblem proximity P (Vr, VQ), again selecting the threshold value ei, according to P (vr, vo) <ei, can be considered represented by an instance of the node vj / f Result of the decision preferably Finally, the process instance of the network, retrieval the right side contains the node Rc ^ esult example represented in the process of network re, if less than the threshold value ei, taken out of the edge node, may be optimized to obtain the decision RMprasult.
2. 如权利要求1所述的两阶段滚齿工艺相似实例检索方法,其特征在于,所述步骤⑴中的工艺实例网络构建过程如下: Stepl计数变量i = 1,j = 1,设计实例变量nextCase = 0; Step2用设计实例物元表示历史加工实例,作为无向加权图中的结点〇j; Step3当j<m,否则转到Step8; Step4 当i<k,nextCase = j + l, 1, 2. The two-stage gear cutting method similar case retrieval claim, wherein said step of process ⑴ example network was constructed as follows: Stepl counter variable i = 1, j = 1, examples of the design variables nextCase = 0; Step2 represents historical example was designed with a processing element instance, as a non-directed weighted graph 〇j the node; Step3 when j <m, otherwise go Step8; Step4 if i <k, nextCase = j + l,
Figure CN103970886BC00031
计算第nextCase个设计实例关于第i个物元特征与第j个设计实例第i个物元特征的接近度p (0 (nextCase) i,Oji),否则转到Step7 ; Step5按照特征权重,计算第nextCase个设计实例物元与第j个设计实例物元的综合接近度P (OnextCase,0j),作为无向加权图中的边θΐί -〈OnextCase,0j〉上的权(nextCase) j ; Step6 i = i+l,转到Step4; Step7 j = j+l,转到Step2; Step8无向加权图的每个结点0 j、对应边ek =〈OnextCase,0 j>、权W (nextCase) j都已得到,组成G =〈0,E,W>,算法结束。 Calculation nextCase th design examples proximity p on the i-th object element wherein the i-th object element wherein the j-th design example (0 (nextCase) i, Oji), otherwise go Step7; Step5 According to a feature weight, calculated the first nextCase th design examples thereof element and the j-th design examples thereof membered integrated proximity P (OnextCase, 0j), as undirected edges θΐί weighting figures - <OnextCase, 0j> right (nextCase) j on; to Step6 i = i + l, go Step4; Step7 j = j + l, go to Step2; Step8 undirected weighted graph each node 0 j, corresponding edge ek = <OnextCase, 0 j>, weight W (nextCase) j are obtained, consisting of G = <0, E, W>, the algorithm ends.
3.如权利要求2所述的两阶段滚齿工艺相似实例检索方法,其特征在于,步骤(1)中所述的滚齿工艺实例库的逻辑结构是实例库的本质描述,是滚齿工艺相似实例检索的基础; 具体构建过程如下: Stepl工艺实例的物元模型描述: 设计实例物元:设计实例是在满足特定设计要求下所获得的设计结果,其物元模型表示为: 3. The two stage according to claim 2 hobbing method similar case retrieval process, wherein, in step (1) in the logical structure of the roller teeth is an essential process instance database library described example, a hobbing process examples of similar base retrieval; in particular constructed as follows: Stepl matter element model describing an example of the process: design examples thereof element: design example is designed to result in meeting specific design requirements obtained, the object-element model represented as:
Figure CN103970886BC00032
其中,N表示设计实例的编号;c表示设计实例的特征名;v是N关于c的量值,记作v = c (N),N,c和v称为物元R的三要素,c和v构成的二元组M= (c,v)表示物元N的特征,称为特征元,根据实例的名称N可以建立实例的索引, 问题实例物元:问题实例是一组约束的集合,它反映了产品在工程语义上的设计要求, 其物元模型表示为: Where, N denotes the number of design examples; wherein c represents the name of the design example; v N value is about c, denoted by c = v (N), N, and v c was referred to as the three elements of the element R, c tuple M = (c, v), and v represents a configuration element N of the features, called feature element may be indexed according to the name of the instance of example N, the problem instance was element: problem instance is a set of constraints it reflects the product on a semantic engineering design requirements, its matter-element model is expressed as:
Figure CN103970886BC00033
其中,No表示问题实例的编号;CQ表示问题实例的特征名;VQ是No关于CQ的量值三元组, 表示为州=〈電1,仰化,¥沉>,1 = 1,2,...,11;¥(^1,¥(^表示特征(^的设计需求区间;¥(^表示特征CQi的权值; Step2构建工艺实例的有限集合KTemp的主要内容ContCase11: 将设计工艺实例用物元描述后,得到每个设计工艺实例的特征元,作为ContCase11的主体,从而将ContCase11表示为{Ni,I,J,matterij,MapIsi | i<m, j<k},其中,m为设计工艺实例个数,k为物元特征个数,依据编号N,依次将设计工艺实例放入ContCase11中,matteriN = (ci,vi) ,cieC,vieV; Step3运用权力要求2中建立的工艺实例网络来表征KTempl二元关系的有限集合Relation =〈0,E,W>,0是工艺实例网络单元中结点的集合,E是边的集合,W是权值的集合; Step4运用二元组TempCase(tc),tc= {KTemp,Relation}描述滚齿工艺实例库的逻辑结构,其中,1 Wherein, No represents the number of problem instance; wherein CQ represents the name of the problem instance; No on the VQ is CQ magnitude triples, represented as state = <1 electrically, of Yang, Shen ¥>, 1 = 1,2, ..., 11; ¥ (^ 1, ¥ (^ represents the characteristics (design ^ demand interval; ¥ (^ represents the weight of the characteristic CQi; main content ContCase11 Step2 build a limited set of examples KTemp process: process design example after matter element is described, wherein each process instance to obtain design element, as a host ContCase11, thereby ContCase11 represented as {Ni, I, J, matterij, MapIsi | i <m, j <k}, where, m is examples of the design process number, k is the element number of the characteristics thereof, according to the number N, the order of the design process instances into ContCase11, matteriN = (ci, vi), cieC, vieV; Step3 use of power requirements of the process of establishing example 2 relation network characterized finite set of binary relations KTempl = <0, E, W>, 0 is set in the process of the example network node element, E is the set of edges, W is a set of weights; Step4 use tuple TempCase (tc), tc = {KTemp, Relation} hobbing process described logical structure of the database instance, where 1 ^_={161^1〇11,(:011忧&8611|11彡〇},了61^1〇11是1^即的代号标示符,(:011忧&8611 = {Νί,Ι, J,matterij,MapIsi | i^im, j^ik} ,Relation = <0,E,ff>〇 ^ _ ^ = {161 1〇11, (: worry 011 & 8611 | 11 billion San}, ^ 61 ^ 1 is 1〇11 i.e. the identifier code, (: worry 011 & 8611 = {Νί, Ι, J, matterij , MapIsi | i ^ im, j ^ ik}, Relation = <0, E, ff> square
4. 如权利要求1所述的两阶段滚齿工艺相似实例检索方法,其特征在于, 所述步骤⑶中的基于表达式驱动的滚齿工艺实例一阶段检索的具体步骤包括, Stepl根据数据字典,针对具体滚齿工艺参数决策问题,输入物元特征包括工件类别、 法向模数、齿数、压力角、螺旋角、材料、精度、径向变位系数、滚切方式的值转化为输出物元特征包括滚刀类别、精度、头数、螺旋升角、滚刀转速、轴向进给速度、径向进给速度、滚切余量、进给量、切削液的值,输出物元特征的值,用问题实例物元表示Rproblani (N〇,0),V〇); Step2根据检索规则,制定检索表达式,具体表示为:Exp= {(CQ1 =滚齿类别,VQ1=(圆柱斜齿轮,圆柱斜齿轮))AND (CQ2 =滚齿精度,VQ2 = (AA,AAA)) AND (C03 =滚齿头数,VQ3 = (1, 2))AND···}; Step3检索滚齿工艺实例库中符合表达式Exp的设计工艺实例,即为候 1 to 4. The two-stage gear cutting method similar case retrieval claim, wherein said step roller ⑶ based on the expression of specific steps the drive teeth process instance includes a phase retrieval, the data dictionary Step1 , hobbing process parameters for a particular decision problem, wherein the input element comprises a workpiece matter category, the normal module, number of teeth, pressure angle and helix angle, the material, the accuracy, the radial displacement coefficient, hobbing mode into an output value thereof characterized in comprising a hob element category, precision, number of heads, helix angle, hob rotational speed, axial feed velocity, the radial feed speed, rolling balance cut, feed rate, the value of the cutting fluid, wherein the output element thereof value, the problem was with the element represented examples Rproblani (N〇, 0), V〇); Step2 according to a search rule to develop search expression, specifically expressed as: Exp = {(CQ1 = hobbing category, VQ1 = (helical gear, cylindrical helical gear)) AND (CQ2 = hobbing precision, VQ2 = (AA, AAA)) AND (C03 = hobbing head number, VQ3 = (1, 2)) AND ···}; Step3 retrieving hobbing examples of process design process repository instance in line with the expression Exp, namely climate 决策实例Rr_it,假定检索出的实例个数为RNum。 Examples decision Rr_it, assume that the number of examples is retrieved RNum.
5. 如权利要求1所述的两阶段滚齿工艺相似实例检索方法,其特征在于, 所述步骤(4)中的基于工艺实例网络的滚齿工艺实例二阶段相似检索的具体步骤包括, Stepl 计数变量i = l,m=l,r=l; Step2 当r<RNum,否则转到Stepll; Step3当i,η为问题物元特征个数,否则转到Step6; 1 to 5. The two-stage gear cutting method similar case retrieval claim, wherein said specific step in step (4) based on the hobbing process example of a process example of a network comprises a two-stage similarity retrieval, Step1 counting variable i = l, m = l, r = l; Step2 when r <RNum, otherwise go Stepll; Step3 when i, η is the element number of the characteristics thereof problem, else go to Step6;
Figure CN103970886BC00041
计算结点所代表的实例第i个物元特征与问题实例第i个物元特征之间的接近度P (vri,v〇1); Step5 i = i+l,转到Step3; Step6按照特征权重,计算结点vr所代表的实例物元和问题实例物元之间的综合接近度P (vr,vo); Step7选定阈值ei,若p (vr,vq) <ei,可认为该结点为优选决策实例,转到Step8,否则转到Step2; Step8若m〈 = NodeNum,NodeNum为结点个数,而且m辛r,计算元优选决策实例结点相连接的结点权值Wrm,m的值在结点总个数范围内,m乒r,与P (Vr,VQ)相减后的绝对值I Wrm-P (Vr, V0) I <ei,则提取该结点为可能优选决策实例,否则,转到Stepl 1; Step9 m=m+l,转到Step8; SteplO r = r+l,转到Step2; Stepll得到优选决策实例和可能优选决策实例就是检索出来的结果,算法结束。 Wherein the i-th element composition represented by the calculated examples of the problem instance node proximity between the i-th element thereof wherein P (vri, v〇1); Step5 i = i + l, go to Step3; Step6 According to a feature weight computing nodes vr represents examples thereof element and problem instance was integrated proximity between the element P (vr, vo); Step7 selected threshold ei, if p (vr, vq) <ei, is considered the junction examples of preferred decision point, to Step8, otherwise go to Step2; Step8 if m <= NodeNum, NodeNum is the number of nodes, and m oct r, node weight calculation element is preferably connected to the decision node Wrm * value, the value of m within the total range of the node number, m ping r, the absolute value of the P (Vr, VQ) subtraction I Wrm-P (Vr, V0) I <ei, is extracted as the node may be preferred examples of decision, otherwise, go to Stepl 1; Step9 m = m + l, go Step8; SteplO r = r + l, go to Step2; Stepll give examples of preferred examples of decisions and decisions may be preferably retrieved result is, ends the algorithm .
CN201410213426.5A 2014-05-20 2014-05-20 A two-stage gear cutting method similar case retrieval CN103970886B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410213426.5A CN103970886B (en) 2014-05-20 2014-05-20 A two-stage gear cutting method similar case retrieval

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410213426.5A CN103970886B (en) 2014-05-20 2014-05-20 A two-stage gear cutting method similar case retrieval

Publications (2)

Publication Number Publication Date
CN103970886A CN103970886A (en) 2014-08-06
CN103970886B true CN103970886B (en) 2017-05-03

Family

ID=51240384

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410213426.5A CN103970886B (en) 2014-05-20 2014-05-20 A two-stage gear cutting method similar case retrieval

Country Status (1)

Country Link
CN (1) CN103970886B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104881530B (en) * 2015-05-18 2017-10-20 重庆大学 Based on dry Qiejia hobbing method for optimizing process parameters

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2000043910A1 (en) * 1999-01-22 2000-07-27 Kent Ridge Digital Labs Method and apparatus for indexing and retrieving images using visual keywords
CN101051306A (en) * 2006-04-07 2007-10-10 同济大学 Punching part sample researching method
CN103207922A (en) * 2013-04-28 2013-07-17 中国兵器工业第五九研究所 Case retrieval method based on precise plastic making database
CN103729525A (en) * 2014-01-26 2014-04-16 重庆大学 Gear hobbing method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2000043910A1 (en) * 1999-01-22 2000-07-27 Kent Ridge Digital Labs Method and apparatus for indexing and retrieving images using visual keywords
CN101051306A (en) * 2006-04-07 2007-10-10 同济大学 Punching part sample researching method
CN103207922A (en) * 2013-04-28 2013-07-17 中国兵器工业第五九研究所 Case retrieval method based on precise plastic making database
CN103729525A (en) * 2014-01-26 2014-04-16 重庆大学 Gear hobbing method

Also Published As

Publication number Publication date
CN103970886A (en) 2014-08-06

Similar Documents

Publication Publication Date Title
Kahraman et al. Fuzzy multi-criteria evaluation of industrial robotic systems
Dereli et al. Optimizing cutting parameters in process planning of prismatic parts by using genetic algorithms
Dweiri et al. Fuzzy surface roughness modeling of CNC down milling of Alumic-79
Sun et al. Soft fuzzy rough sets and its application in decision making
Ahilan et al. Modeling and prediction of machining quality in CNC turning process using intelligent hybrid decision making tools
Ju et al. Extension of VIKOR method for multi-criteria group decision making problem with linguistic information
US7065420B1 (en) Integrated real-time feature based costing
Carlsson et al. Multiobjective linguistic optimization
Lau et al. A fuzzy logic approach to forecast energy consumption change in a manufacturing system
Own et al. Forecasting fuzzy time series on a heuristic high-order model
Fenggou et al. The study of high efficiency and intelligent optimization system in EDM sinking process
Zhang A rough set approach to intuitionistic fuzzy soft set based decision making
Kovac et al. Application of fuzzy logic and regression analysis for modeling surface roughness in face milliing
Ahmad et al. Current trend in computer aided process planning
US9519859B2 (en) Deep structured semantic model produced using click-through data
Gajate et al. Tool wear monitoring using neuro-fuzzy techniques: a comparative study in a turning process
Cakir et al. An expert system approach for die and mold making operations
Jiang et al. An automatic process planning system for the quick generation of manufacturing process plans directly from CAD drawings
EP0506038A2 (en) Processing apparatus having rule base
CN103235556B (en) Feature-based method of manufacturing complex parts NC
SMITH et al. Process plan generation for sheet metal parts using an integrated feature-based expert system approach
Athawale et al. A TOPSIS method-based approach to machine tool selection
Pusavec et al. Sustainable machining of high temperature Nickel alloy–Inconel 718: part 2–chip breakability and optimization
Rao et al. Flexible manufacturing system selection using a combinatorial mathematics-based decision-making method
Armaghan et al. An application of multi-criteria decision aids models for case-based reasoning

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
GR01