885191353基于健壮性的机械设计方法中英文资料

《885191353基于健壮性的机械设计方法中英文资料》由会员分享，可在线阅读，更多相关《885191353基于健壮性的机械设计方法中英文资料（15页珍藏版）》请在装配图网上搜索。

1、基于健壮性的机械设计方法Abstract: The concept of robustness derived from the field of cont role Engineering and cont role theory, and the concept of robust design put for ward by Taguch and used in quality engineering get into the area of mechanical engineering in recent years. In the is paper we discuss four a

2、spects of the problem: robust design based on quality engineering; robust design based on ensitivity analysis; robust structure; structure. control synthesis robust design. Finally, the development in each aspect is pointed out, and some conclusions are drawn.Keyword: Strong design; Quantity enginee

3、ring; An analysis; The structure control integral whole turn Lead the speech摘要: 健壮性的概念起源于控制工程的控制理论, 并习称鲁棒性。而后由于田口提出健壮设计的概念而引入质量工程, 近年来逐渐进入机械工程领域。本文从基于质量工程的健壮设计, 基于灵敏度分析的健壮设计, 健壮结构, 结构.控制一体化的健壮设计等四个方面概述了其发展, 最后提出尚待研究的机械健壮性设计的一些关键问题。关键词: 健壮设计; 质量工程; 灵敏度分析; 结构控制一体化引言在机械系统设计完成而进入制造和使用阶段时, 由于一系列不确定因素的影响,

4、 可能会出现性能上大的变化, 甚至造成产品失效, 从而导致质量不稳定。如何使设计出的产品具有健壮性, 亦即使产品和过程具有抵抗外界干扰所引起的质量波动的能力, 是目前设计方法的前沿课题。由此提出了机械健壮性设计的概念。1基于质量工程的健壮设计1. 1质量工程质量工程(Q E) 是对产品开发全过程质量保证和质量控制的系统工程方法, 包括设计总质量工程和制造质量工程。设计质量工程主要有: 健壮性设计; 可靠性设计; 满足产品全面质量指标的计算机集成化智能设计; 以健壮性、可靠性为重点的全面质量管理和质量保证技术。制造质量工程主要解决设计质量的符合性问题。1. 2健壮性设计的概念所谓健壮性设计, 就

5、是赋予产品或过程健壮性、高性能和低成本的设计。它是一种性能、质量和成本综合的功能优化设计方法, 是一种着眼于经济效益, 立足于工程技术的质量设计和管理技术。这种设计方法是在传统的工程设计方法上发展而来的, 其发展过程为: 数学优化工程优化(试验设计) 田口方法(Tagnch imethod) 健壮设计。健壮设计能解决大规模产品设计问题。它加强了田口方法的系统设计, 突出了QDF (质量功能展开) 的先导作用。1. 3健壮性设计方法的改进(1) 不确定因素的不同形式: 田口方法仅考虑了随机不确定因素。文献1 讨论了其它形式的不确定因素, 并指出这些因素对优化目标及质量信噪比有影响, 从而改进了S

6、.N 表达式。(2) 其它的搜索方法: 田口方法仅仅用试验方法在整个设计空间中进行搜索, 经过一系列的正交试验, 可以选出一组最优的设计参数。田口方法使用的是两参数试验设计方法, 其改进方法: 使用二进式搜索; 在设计空间使用爬山法; 还可使用随机搜索方法等。(3) 有约束条件的设计: 田口方法是在无约束条件下使用的, 如果存在约束, 可利用约束搜索方法, 将原约束转换为概率形式。另一种方法是使用定义边界的方法, 定一个分段函数。还可使用Monte2Carlo 法求解。文献2 指出了田口方法在处理约束及多目标问题方面的不足, 发展了田口方法, 使它可以处理约束及多目标的问题。文中使用统计分析A

7、NOVA 的概念来确定优化目标函数中各因素的重要性, 排除了不重要因素的影响, 得出唯一的可行最优解。文献3 成功地利用单调性分析技术将有约束的健壮性设计问题简化为无约束优化问题, 以此来减小计算量。文献4 提出一个基于单调性分析的方法, 来解决或帮助解决以下问题: 找出设计变量x 的值, 使不可控的任意参数F对性能特性y= f (x, F)的影响最小, 即对不可控因子的变异的敏感性最小, 以此完成健壮设计。这种方法与田口方法相比在于它不需要统计试验, 而是基于性能特征函数y= f (x,F) 的单调性。1. 4基于质量工程的健壮设计文献5, 6 提出了一种健壮设计的途径: 极小化由噪声因子(

8、不可控因子) 和可控因子引起的变异。健壮设计的基本原理是通过使变异造成的效果最小而不必消除其产生的来源, 以此来提高产品质量。由此可把问题分为两大类: 类: 将噪声因子(不可控参数) 的变异产生的对性能的影响减小到最低限度。 类: 将可控因子(设计变量) 的变异产生的对性能的影响减小到最低限度。ChenW ei 指出田口的“参数设计”是由第 类问题发展而来, 并提出解决、 两类健壮设计问题的过程包括以下三个主要步骤:(1) 运用响应面法, 建立与所有重要的可控和噪声因子有关的响应面模型。(2) 针对健壮设计问题的不同类型, 推导响应的均值和方差的函数。(3) 运用相容决策支持问题(DSP) 解

9、法, 求得健壮设计解。ChenW ei 提出的这种方法具有以下优点: 可以引入工程约束, 对多个设计目标之间的均衡有很大的灵活性; 可以用响应面模型来反映出耦合和非线性的效果, 比田口的线性模型更精确; 响应面模型可作为快速分析模块应用于不同类型的健壮设计问题中。但这种方法还有一些不足: 目标函数和约束方程是用试验设的统计和特殊的二次模型来进行近似的, 对于高度非线性模型和二阶模型应用仍不够满意; 在公式推导中用到了一些假设, 影响模型的精确性。Park inson7 9 将工程模型的健壮性分为两类: 可行度健壮性(确保设计约束即使在有变异的情况也能满足) 和灵敏度健壮性(其目的是要减少设计对

10、变异的灵敏度) , 并用一优化问题来表达定量设计问题。数学模型为M in f i (x , p ) i = 1, 2, ., qs. t. g j (x , p ) bjj = 1, 2, ., q (1)式中x 为设计变量(其值在上下限之间) 且是可控制变量;p 为不可控制参数, 如误差等。Park inson 指出健壮设计的核心是一个优化问题, 健壮设计的一个关键概念是变量和参数的变异传递给了设计函数, 使目标和约束都产生变异。变量和参数的统计分布导致设计函数的统计分布。健壮设计的目的是减小所诱发的变异。针对工程模型,讨论了下列七种健壮设计方法: (1) 传递变异法； ( 2) 优化后分析

11、；(3) 参数约束；(4) 公差盒方法; (5) 随机优化; (6) 田口方法; (7) 相关的其它方法。对于每种方法的优缺点用实例进行了比较。近期还有一些其它发展: 调节参数法; 引入变量之间的相关性; 处理非正态分布参数的方法; 其它形式的不确定性。2基于灵敏度分析的健壮设计文献 10 研究了优化问题中参数的灵敏度, 提出了灵敏度导数的概念。从数学上讲, 灵敏度导数是指目标函数和设计变量对所研究的参数求导。Jaro slaw 10 探讨了如何由一个约束优化问题获得灵敏度导数(灵敏度方程) 的等式,还探讨了灵敏度方程可解性等问题, 最后总结了灵敏度分析的用途: 目标函数和设计变量的向外插值;

12、 多目标优化;预测约束状态的变化;把大的优化问题分解为一些子问题。文献 11 在前文的基础上, 研究了机构系统的性能质量和容差灵敏度, 提出系统性能质量对容差灵敏度的确定方法和判定对容差最不敏感的健壮设计的通用方法。使用该方法, 可以确定出受尺寸容差影响最小的健壮机构, 同时可得高性能质量的闭合形式的理想容差分布。文献 12 研究了通过极小化灵敏度来实现健壮设计。其基本思想是: 在设计阶段就通过选择适当的变量, 使产品的质量对不确定性因素的敏感性最小, 使产品具有健壮性。这种方法的另一优点是能设计出允许更大容差的产品而同时具有较低的成本。Belegundu11 首先定义了两类变量: 设计变量b

13、 和不确定变量x , 将机械或结构的响应F 表达为F = F (b, x ) (2)由此推导出基于最小灵敏度的健壮性设计问题的格式。文中的方法没有假设变量的概率分布, 但提出的两个定理, 证明了如果不确定变量考虑成为具有概率分布的随机变量,那么灵敏度的降低就意味着失效概率的减小。其结论为: 灵敏度越小, 安全性越高。3对健壮结构的研究田口在1986 年从质量工程的角度提出健壮设计这一概念以后, 机械工程领域的学者们就试图将健壮性的概念引入机械工程领域。荒川雅生等13 15 研究了基于灵敏度分析的结构优化设计, 提出了定性灵敏度和定性优化设计的概念, 给出了定位优化设计的通用算法13 。尔后又将

14、这种算法推广到多目标优化, 并义了三种模糊语言“小”,“标准”,“大”, 为它们配置了模糊子集, 子集中包含了离散化的设计变量。文献 14 利用文献 13 的结论, 提出一种利用已知的简单优化设计的结论来求解复杂问题的定性灵敏度的方法。文献15 提出了应用试验结果和定性灵敏度来对实际的设计对象进行定性优化设计方法, 使设计结果具有健壮性。1991 年, 文献 16 第一次提出了健壮结构的概念, 以后又连续发表文章, 对健壮结构作了较深入的探讨。他们提出的健壮结构的定义是: 不管结构内部参数还是外部环境的变动, 都对结构的影响达到最小, 并提出了健壮结构设计问题的格式。文献 17 发展了健壮结构

15、的概念, 使其从狭义扩展到广义, 并列举了在广义的健壮结构中, 一些目标函数可能的取法。还应用广义的健壮结构概念, 进行了算例分析, 得出的结论为: 广义健壮结构用于静态问题中的大部分问题都很有效, 但些性能指标的效果欠佳。同时提出以后的工作重点是寻找更好的性能指标作为目标函数。文献 18 将健壮结构从静态延伸到动态, 提出了广义的动态健壮结构的概念和广义动态健壮结构的设计方法。其要点是利用现有的有限元分析技术, 以结构自由振动的特征值作为目标函数, 对其进行灵敏度极小化分析, 从而求出健壮性参数。文献19, 20 讨论了模糊健壮结构, 指出以前都是以灵敏度的方法来设计健壮结构的, 但灵敏度仅

16、仅提供了局部信息, 因此不能确定用灵敏度设计出的结构在大的范围内具有健壮性。在具体的解法中, 使用了一个隶属函数Lij, 并用算例讨论了隶属函数的变化造成的影响。4对结构.控制一体化的健壮性研究现代机械工程技术发展到今天的水平, 已不再局限于某一领域。作为结构和控制一体化的智能结构, 已广泛应用于结构的主动控制、机器人、航天、计算机硬盘等重要领域,成为新的研究热点。所以对于结构.控制一体化的健壮化性设计的研究, 具有重要意义。长期以来, 结构和控制部分都是作为单独的模块来设计的, 结构经过优化设计后, 再将最优控制系统加到结构上。这样导致的结果是无法考虑两个系统的耦合, 使得系统参数间相互干扰

17、, 设计出的结果可能效果很差。所以有必要将结构和控制集成后再进行设计。文献 21 讨论了结构和控制的集成设计, 设计的对象是一具有主动控制的结构, 约束施加于闭环系统的阻尼参数上。这一问题被定义为多目标优化, 以结构重量和控制系统的能量作为目标函数, 以结构件的横截面积作为设计变量。多目标优化问题的难点在于几乎不存在一个解使所有目标函数同时达到最优。所以文中使用了可行最优解的概念。求解多目标优化问题一般的是用加权因子将各目标线性组合到一个目标中, 但这样会导致求出的解不一定是全局最优解。文中采用了博弈论的方法来求解、使用相关权值来处理多目标间的关系, 同时进行了灵敏度分析, 以保证结果的健壮性

18、。文献22 较深入地讨论了具有主动控制的结构的健壮设计问题, 指出柔性结构从本质上讲是分布参数系统。有限元计算中的近似处理会导致参数的误差, 并会导致模态的丧失。结构参数中也存在着相当的不确定性, 这些不确定性会明显地影响控制系统的稳定性和性能特性。因此必须确定参数扰动对系统性能的影响。反馈控制系统的鲁棒性成为设计中的一个重要问题。在设计过程中需考虑两种鲁棒性: 稳定性鲁棒性和性能鲁棒性。以往对稳定性鲁棒性的研究主要在频域中用单值分解来完成, 而对性能鲁棒性的研究则用灵敏度分析在时域进行。同时在设计过程中研究稳定性鲁棒性和性能鲁棒性的文献较少。文献22 同时考虑这两种鲁棒性, 用闭环系统特征值

19、矩阵的状态数来反映稳定性鲁棒性, 以相关灵敏度来示性能鲁棒性。文中为了使系统达到健壮最优, 采用了多目标优化, 目标函数有4 个: (1) 稳定性鲁棒性; (2) 性能鲁棒性; (3) 结构重量; (4) 控制系统能量。采用协同模糊运筹学求解问题的可行最优解。5结束语(1) 基于质量工程的健壮性设计, 可以称之为“广义”的健壮性设计方法, 主要是在设计中对机械结构在制造中产生的偏差加以克服, 使产品具有抵抗制造误差干扰的能力。提供的方法大都是以试验设计技术为基础的, 而且考虑的均为静态问题, 不考虑产品在工作过程中本身的性能。今后主要工作在于提出更有效的健壮设计方法。( 2) 基于灵敏度分析的

20、健壮性设计, 可以称之为“狭义”的健壮性设计方法, 主要是对优化以后的结果进行分析, 使其具有健壮性。适用面较窄。(3) 目前结构.控制一体化的健壮性研究仅仅考虑到集成后控制系统的鲁棒性, 而对于结构参数的健壮性未做研究, 所以集成后的系统健壮性是一个新的且有价值的课题。(4) 健壮性的概念起源于控制(鲁棒性) , 后逐渐引入质量工程, 现已向机械工程渗透。机械系统的健壮性设计是一门崭新的学科, 它将对设计方法的更新产生巨大的影响。尤其是在当前各学科互相渗透、互相补充的情况下, 机械的健壮性和结构.控制一体化健壮设计是这一领域的前沿课题。如何建立健壮性的评价指标, 是重要的基础研究; 如何把控

21、制中的鲁棒性和机械结构的健壮性有机地结合起来,是理论研究的重点所在。这也将对智能结构的广泛应用起到很大的推进作用。According to the machine design method of the halenessAbstract: The concept of robustness derived from the field of cont role Engineering and cont role theory, and the concept of robust design put for ward by Taguch and used in quality enginee

22、ring get into the area of mechanical engineering in recent years. In this paper we discuss four aspects of the problem: robust design based on quality engineering; robust design based on ensitivity analysis; robust structure; structure. control synthesis robust design. Finally, the development in ea

23、ch aspect is pointed out, and some conclusions are drawn.Keyword: Strong design; Quantity engineering; An analysis; The structure control integral whole turn ； Lead the speechAt the machine system design completes but enters the manufacturing and uses the stage, because of The influence of a series

24、of indetermination factor, may appear the function up the big variety, very Go to make product expired, thus cause the quantity unsteady .How make design to produce The article has the haleness, that is make the product and processes have the outside interference of holdout leadQuantity of rise the

25、ability of the motion, design the front of method to follow the topic current. Lift from here The machine concept of the strong sex design.1 according to strong design of the quantity engineering1. One quantity engineeringThe quantity engineering( the Q E) develops the whole process quality control

26、and quantities to the product The system engineering method of the control, include to design the total quantity engineering and the manufacturing quantity works Distance. Design the quantity engineering to mainly have: Strong sex design; Dependable sex design; Satisfy to produceThe article calculat

27、or of the overall quantity index sign gather turn the intelligence design; With the haleness, credibility Sex manages for the overall quantity of the point and quality control technique. Make the quantity engineering Mainly resolve to design the quantity to match the sex problem.1. 2 concepts of the

28、 strong sex designsSo-called and strong sex design, is to give the product or the process haleness, high performances And low cost design It is the function that a kind of function, quantity and costs synthesize excellent。Turn to design the method, is a kind of to fix attention on in the economic pe

29、rformance, have a foothold in the technical quality of engineering Quantity design and management technique .This kind of design method designs the square in the traditional engineering Development on the method since then of, it develop process is: Mathematics is excellent to turn the engineering i

30、s excellent to turn( try Check the design) a method( the Tagnch imethod) of farmland strong design.The strong design can resolve the large-scale product design problem. It strengthened the farmlandThe system design of the method, outstanding the QDF( the quantity function launch) leads to make first

31、 se.1. 3 halenesses design the improvement of method(1) Different form of the indetermination factor: A method of farmland considered only random Indetermination factor. Cultural heritage1 discussed theindetermination factor of other forms, and point A these factors to excellent turn the target and

32、the quantity letter to compare to have the influence, thus improved S. The N expression type.(2) Other manhunt method: A method of farmland uses to experiment the method only at whole The design carries on the manhunt in the space, experimenting through just hand over of a series, can select Superio

33、r design parameter. What a method of farmland use is 23 the numbers experiment the designMethod, it improves the method: Use two enter the type manhunt; At design the space usage to climb a mountainMethod; Can also use the random manhunt method etc.(3) Control the conditional design: A method of far

34、mland is under the condition of have no stipulation Usage of, if the existence control, can make use of to control to search the method, will control to turn at first Change for all rate form. Another method is an usage definition the method of the boundary, settle a Divide a function. Can also use

35、the Monte2 Carlo method to solve. Cultural heritage2 pointed out a method of farmland at handle to control and many target problem squareShortage of face, developped a method of farmland, make it be able to handle to control and many target of Problem.The usage in text statisticses the analytical AN

36、OVA concept to certain excellent turn the target The importance of each factor in the function, expeled the influence of the of no account factor, get only An of can go the superior solution.Cultural heritage3 make use of monotonous sex analysis technique and will have the haleness of the stipulatio

37、n successfully Sex design problem simplification for have no control excellent turn the problem, let up the amount of calculation with this. TextDedicate4 put forward an according to analytical method of monotone, solve or help to solve Following problem: Find out the design changes the value of mea

38、sure the x, make cant control arbitrarily the parameter F toThe function characteristic y= the f(x, F) influence is minimum, namely to cant control the variation of factor ofThe sensitivity is minimum, completing the strong design with this. This kind of method and farmland methods compare Lie in it

39、 and dont need to statistics to experiment, but according to the function characteristic function y= f( x,F) Of monotonous sex.1. 4 according to strong design of the quantity engineeringCultural heritage5,6 put forward a kind of path of strong design: Turn smallest from the voice of Factor( cant con

40、trol the factor) and can control the variation that factor cause. the strong design The origin reason is a result that passes to make variation result in minimum but need not remove what it produce Source, raise the product quantity with this. Can is divided into the two major type to the problem fr

41、om here:Is output the variation of a factor (cant control the parameter) to functionOf influence let up the lowest limit: Will be able to control the variation of the factor(the design changes the quantity) output to function ofThe influence lets up the lowest limit.The ChenWei point out that oscula

42、r the parameter design of farmland is from the problem hair The exhibition is since then, and put forward process that resolves two strong design problems of include withNext three and main step:(1) The usage responds to face the method, establishment and all important can control and voice ，because

43、 ofThe son is relevant of respond to and face model.(2) Aim at the different type of strong design problem, deduce to respond to of all be worth And function that square differ.(3) Make use of the compatible decision support problem( DSP) solution method, beg the haleness to establish Account the so

44、lution.This kind of method that the ChenWei put forward has the following advantage: Can lead into the work The distance control, designing the well balanced between target to have to many very greatly vivid; Can Use to respond to face the model to reflect to match and not line result, compare the l

45、ine of the farmland Sex model is more accurate; Responding to face model can be the fast nalytical mold piece apply in not In the strong design problem of the of the same kind type.But this kind of method have some the shortage: TargetThe function peace treaty tie the square distance is to use to ex

46、periment the covariance of design and special two times model Carry on look like of, for the height not the line model and two rank models the application is still not enough Be satisfied with; Deduce in the formula convenient arrived some assumptions, accurate sex of the influence model. The Park i

47、nson7-9 is divided into the haleness of the engineering model two type: Can the degree of the line Haleness and work properly A haleness, counteract An excellent turn the problem to express the fixed amount design problem. Mathematics model isM in f i (x , p ) i = 1, 2, ., qs. t. The g j( x, p) bj j

48、=1,2,., q (1)The x in type changes the quantity( it is worth between the top and bottom limit) for the design and is to control to change the quantity; P for cant the control parameter, if error margin etc. The Park in son points out that the core of strong design is an excellent turn the problem, s

49、trong The variation that a key concept design is to change the quantity and parameters deliver to the design letter Number, make the target peace treaty tie produce all variation. The covariance that changes the quantity and parameters distributes to cause The covariance of the design function distr

50、ibute. The purpose of the strong design is to let up induct of change Different.Aim at the engineering model, the Park inson discussed following seven kinds of strong design squaresesMethod: (1) Deliver the variation method( M ethod of T ransm it ted V ariat ion, M TV ) ; ( 2) Excellent turn analyti

51、cal behind( Po st Op t imality A nalysis- POA ) ; (3) The parameter control( Paramet ric Const rains- PC) ; (4) Mr. A method( To lerance Box App roach- TBA ) ; (5) Random excellent turn (Stochast icOp tm izat ion - SO ) ; (6) A method of farmland; (7) Related Other method. Use for the merit and shor

52、tcoming of each method the solid example carried on comparison.Have some other developments in the near future: Regulate the parameter method; Lead to go in to become Relativity; Processing not the positive distributes the method of the parameter; The indetermination of other formsSex.2 according to

53、 an analytical strong designCultural heritage10 studied excellent turn the problem in the intelligent degree of the parameter, put forward to work properly A concept that leads the number. Speak from mathematics, the intelligent degree lead the number to mean the target function and The design chang

54、es the quantity to beg to lead to the parameter study. The Jaro slaw10 inquiried into how from A control excellent turn the equation that the problem acquires the intelligent degree to lead the number( a square distance), Still inquirying into a square distance can solve sex etc. problem, tally up f

55、or a cent finallyThe use of The target function and designs change the outwardly put of the quantity value; Many targets are excellent to turn;Predict the variety of control the appearance; Greatly excellent turn the problem decomposition as some sub- problems.Cultural heritage11 fore the foundation

56、 of the text up, studied the function quality of the organization system Quantity and permit the bad intelligent degree, put forward the system function quantity to permit bad assurance of the intelligent degree Method and judge in general use method to permit the bad least impressionable and strong

57、 design. Usage That method, can make sure that be subjected to the strong organization that the size permits bad influence least, at the same time Can acquire the high performance quantity to shut the ideal of match the form to permit to differ to distribute. Cultural heritage12 studies to pass to t

58、urn the intelligent degree smallest to carry out the strong design. Its basic thought is: At design the stage to pass the choice to change the quantity appropriately, make the product Of quantity is minimum to the sensitivity of the indetermination factor, make the product have the haleness. This ki

59、nd of method of another an advantage is to design to allow larger permit the bad product but together The hour has the lower cost.Belegundu11 defined first two the types change the quantity: The design changes to measure the b and not indeedChange to measure the x certainly, respond to machine or st

60、ructures expression of F isF = F (b, x ) (2)Deduce from here to design the format of problem according to the haleness of the minimum and intelligent degree. Text The medium method did not suppose to change the quantity all the rate distribute, but two axiomses that put forward, certificateUnderstan

61、d if the indetermination changes the quantity consideration to become have all the rate distribute of change the quantity random,The so intelligent degree lowers and then mean the expiration all the rate let up.Its conclusion is: Work properlyThe degree of is more small, the safety is more high.3 ri

62、ghtnesses of strong and structural researchesThe farmland puts forward the strong design from the angle of the quantity engineering in 1986 thisConcept later, the scholars of the mechanical engineering realm try the concept of the haleness Lead into the mechanical engineering realm. The livings etc.

63、13-15 studied is excellent to turn according to an analytical structureDesign, put forward certainly intelligent degree of sex and certainly sex is excellent to turn the concept of design, give settleIs excellent to turn the in general use calculate way of design13.Expand this kind of calculate way

64、again thereafter many eyesThe mark is excellent to turn, and defined three kinds of misty languages small, the no rmal, Large, installed the misty son to gather for them, the son concentration included long-lost turn of The design changes the quantity.Cultural heritage14 makes use of the cultural he

65、ritage13 conclusion, put forward a kind of make use of to have already know of Simple excellent turn the conclusion of design to solve the complicated problem certainly sex the square of the intelligent degree Method. Cultural heritage15 puts forward the application to experiment the result and certainly intelligent degree of sex come to actual Of design object carry