外文翻譯--用于塑料注射模具設(shè)計和生產(chǎn)的自動基準尺寸-18頁[英文為PDF]【中英文文獻譯文】

外文翻譯--用于塑料注射模具設(shè)計和生產(chǎn)的自動基準尺寸-18頁[英文為PDF]【中英文文獻譯文】,英文為PDF,中英文文獻譯文,外文,翻譯,用于,塑料,注射,模具設(shè)計,以及,生產(chǎn),出產(chǎn),自動,基準,尺寸,18,英文,pdf,中英文,文獻,譯文 用于塑料注射模具設(shè)計和生產(chǎn)的自動基準尺寸 摘要：基準尺寸（或者坐標尺寸）技術(shù)在表面有大量必須有規(guī)定的孔特征的注射模具畫圖方面應(yīng)用非常廣泛。盡管商業(yè)CAD/CAM系統(tǒng)提供了半自動工具來協(xié)助設(shè)計者進行尺寸確定，但它仍然是個非常復(fù)雜的過程。作為使用者，不得不進行規(guī)定每個尺寸標記的位置。這篇文章講的是一種最優(yōu)的尺寸標記的全自動方法。這種方法采用的是動態(tài)工程技術(shù)，在尊重用戶所選擇的標準前提下，使尺寸確定最優(yōu)化。這種方法已經(jīng)作為一種工具插到了商業(yè)CAD/CAM系統(tǒng)中，并且給出了一些實例來說明這個技術(shù)的重要特征。 關(guān)鍵字：自動尺寸 基準尺寸 動態(tài)工程 最優(yōu)尺寸 坐標尺寸 1介紹 CAD/CAM系統(tǒng)現(xiàn)在在塑料模具注射制造業(yè)應(yīng)用非常廣泛，許多公司可用一種實體模擬系統(tǒng)來設(shè)計注射模具。他們用CAD系統(tǒng)來設(shè)計，不僅僅是核心和孔插入到模具（他們是形成模具最重要部分），同時也是在模具裝配中其他所有部分。伴隨著互聯(lián)網(wǎng)技術(shù)的進步和最近CAD網(wǎng)絡(luò)通信的發(fā)展，注射模具的設(shè)計信息可以在產(chǎn)品工程師（設(shè)計塑料部件）和刀具工程師（設(shè)計注射模具），甚至他們位于世界上不同區(qū)域。在設(shè)計信息通過電子方式在產(chǎn)品設(shè)計和刀具設(shè)計中可以有效的交流，制造信息通信在電子和傳統(tǒng)技術(shù)共同作用下在車間完成。數(shù)控機床刀具軌跡或者檢查指令可以直接從CAD/CAM和互聯(lián)網(wǎng)下載給數(shù)控控制器進行加工或者進行檢查工作。然而，對于一個專門機器來說建立一個說明書或許在工程圖紙中規(guī)定。另外，不是所有的加工任務(wù)都是在數(shù)控機床上完成。出于成本預(yù)算的考慮，一些傳統(tǒng)機床，例如鉆床和磨床等，可以方便用傳統(tǒng)刀具完成。通常工程圖紙也在車間里的工程信息通信中扮演重要角色。正交投影工程圖紙可以從CAD模型中自動產(chǎn)生。零件尺寸自動工具同樣由商業(yè)CAD系統(tǒng)提供。然而，就像有chen指出的，那些尺寸工具不能夠根據(jù)圖紙標準和工廠通常采用的工程方法來生成尺寸。 在注射模具的特殊要求中，孔特征的基準尺寸（或坐標尺寸）應(yīng)用得非常廣泛。圖1展示了一張在一個模具制造公司的工廠中可以找到的典型詳細的圖紙。在圖中表明了孔特征和用來規(guī)定這些孔位置的基準尺寸?？梢园l(fā)現(xiàn)這些尺寸顯得非常擁擠，并且人工的來確定這些基準尺寸的位置是非常繁重的工作。最終這張滿帶尺寸的圖紙的質(zhì)量很大程度上依賴于這張圖紙繪圖員的經(jīng)驗。這項研究的目的是發(fā)明一種能夠從一個給定的注射模具的零件上自動的產(chǎn)生基準尺寸。尺寸結(jié)果必須滿足兩個要求：第一，任意兩個尺寸標記不能夠重疊；第二，尺寸標記必須盡可能的接近被測特征。這個問題的關(guān)鍵是研究一種使基準尺寸位置的最優(yōu)化的方法。 2.相關(guān)工作 在規(guī)定機械零件或組裝特征的大小或位置信息中尺寸標記和公差分析是兩個非常接近的工作，并且大多數(shù)的研究工作都集中于公差上。公差的主要研究問題是顯示、分析和合成。公差顯示與把公差信息合并產(chǎn)品模型草圖中相關(guān)。實例中包括由Requicha發(fā)明的實體偏移法、Turner的可行性空間法和Desrocher和Dlement的TTRS法。在Roy和Yu的文章中有更加詳細的觀點。公差分析的目的是確定公差中決定零件公差的合并效果。它被用來證實已知或猜想的給定設(shè)計單個零件尺寸變化的功能性。公差分析的技術(shù)實例包括Monte，Carlo總結(jié)和指出的線性方法。它的主要觀點是綜合合成或公差分配，它是用來確定基于給定裝配的功能性要求的零件公差。最近，Is Lam同時發(fā)表了一個工程方法來解決這個問題。從工程性考慮，根據(jù)不同用戶需求和技術(shù)需求在系統(tǒng)的分析功能性需求基礎(chǔ)上，提出了一種解決尺寸需求的方法。FDT軟件也是用來支持實現(xiàn)這種方法。FDT向功能需求或尺寸基礎(chǔ)中提供用來代表功能需求、尺寸、公差或過程能力的工具。在這個基礎(chǔ)上，這種獲取的函數(shù)方程被分成若干組，并且每組都用來解決所涉及到的函數(shù)需求和公差問題的專一方法策略。在公差公差分析和合并中更多詳細的觀點可以在Roy、Ngoi、Ong，Hong和Chang的文章中找到。 從一個CAD模型中自動的產(chǎn)生尺寸的方法已經(jīng)被提出。在立體幾何學(xué)實體模型技術(shù)上的零件自動顯示尺寸上Yuen做過一些早期的嘗試。從實體模型上提取一些來自二維表面和剪切剪切圓柱上的點。這些點的坐標被安排在一個樹型結(jié)構(gòu)上來產(chǎn)生線性尺寸，并刊登了一個為直徑和半徑尺寸的簡單例子。其他早期的自動尺寸方面的工作由Yu總結(jié)了。最近，Chen發(fā)表了一個更加在自動尺寸方面具有深度的研究。他們的方法是分析多余尺寸，對稱特性的尺寸草圖。，選擇適當(dāng)?shù)囊暯莵硪?guī)定尺寸，并且用一種專業(yè)級系統(tǒng)的方法來確定尺寸定位。這個專業(yè)級系統(tǒng)分析被測特征的幾何形狀和布局，并且確定一個合適的位置來放置一個基于一套規(guī)則的相關(guān)被測特征的尺寸。在一個尺寸位置完成后，構(gòu)造一個禁區(qū)使所有隨后的尺寸不會放在此區(qū)域中。這就避免了兩個尺寸的重疊或交叉。 現(xiàn)存的尺寸放置方法是有限的，取決于這個方法的自然連續(xù)性。例如，在Chen的方法中，特征的測量是被優(yōu)先考慮的，但尺寸的位置是隨后考慮的。這種方法不適合確定坐標尺寸的位置，特別是在注射模具表面尺寸非常擁擠的情況下。這是因為一個基準尺寸的位置影響到其他尺寸的位置。這篇文章講述了我們在解決坐標尺寸位置方面的工作。使用了動態(tài)程序方法使其最優(yōu)化，這種新的方法克服了在現(xiàn)存方法中連續(xù)方法的限制。 3.在基準尺寸中，特征位置通過特征的參考位置和參考基準的水平和垂直距離來確定?；鶞食叽绲娜笔⌒问皆趫D2a中表示。當(dāng)被測兩特征的垂直距離小于尺寸標記大?。ǔ叽缥谋靖叨鹊暮秃拖噜弮沙叽缥谋镜淖钚】臻g）,圖2b中的被選形式就需要了。為了防止不重疊，這個尺寸標記則從缺省位置向上或向下轉(zhuǎn)換。就像圖2c中，尺寸標記的轉(zhuǎn)換是尺寸的單一延長線打斷成了三個部分：被一個傾斜部分相連的兩個水平部分。通過三個參數(shù)使尺寸標記的范圍能夠變換調(diào)整：（i）傾斜部分和尺寸線的水平部分的折角α；（ii）尺寸文本和零件邊緣距離m；（iii）特征的位置和最低位置由下式給出： = = （1） 4．自動基準尺寸 自動基準尺寸的目的是尋找一種使每個基準尺寸位置都達到最優(yōu)的方法。每個過程包括兩個階段：準備階段和最優(yōu)化階段。在準備階段，使最優(yōu)化過程得到簡化的主要參數(shù)將被建立。所有特征使用給出的折角、邊緣偏移和尺寸標記大小將會進行尺寸位置的可行化測試。在最優(yōu)化階段，使用的是動態(tài)程序方法。尺寸標記位置最優(yōu)化是在最重不同的基準系列，包括從他們的缺省位置中每個尺寸轉(zhuǎn)移的最小量，或者從缺省形式使用的盡可能多的最大量。 4.1準備階段 被測特征首先被分成一個或更多的特征系列。對于一個特征系列的每一個特征來說，在此特征系列中至少存在另一個特征以便使他們之間的垂直距離小于尺寸標記的大小。換句話說，在一個特征系列中的所有特征，沒有重疊的情況下在相鄰兩尺寸標記不能用缺省的專用形式測量。相反，至多一個特征能用一個叫做尺寸塊的特征相關(guān)聯(lián)。尺寸快的構(gòu)造涉及到每個在尺寸塊中的基準尺寸的形式和位置。對于一個尺寸塊的每個位置，它的構(gòu)造是唯一確定的。圖3表示在兩個構(gòu)造中的兩個特征系列和它們的尺寸塊 定義1：構(gòu)造的正確性。假如在一個尺寸塊中任意兩個尺寸標記間沒有重疊，并且每個尺寸標記位于它們的末端位置，則這個尺寸塊的構(gòu)造就是正確的。 在圖3b中表示的尺寸塊就是正確的。圖4表示的兩個尺寸塊是不正確的。因為在圖4a中的尺寸標記有重疊所以不正確，圖4b的構(gòu)造中，14.00尺寸標記的延伸線的位置太高，而尺寸要求的位置標記在它之下。 定義2：構(gòu)造末端。有兩種構(gòu)造末端：最高構(gòu)造和最低構(gòu)造。假如一個尺寸是正確的，并且任何再高點（或低點）的位置都是不正確的構(gòu)造則稱此尺寸塊是最高（或最低）構(gòu)造。尺寸塊的末端構(gòu)造取決于和。 圖5a里的構(gòu)造處于它的最高構(gòu)造。因為29.5是它的最高位置，它不能再向上移動了。圖5b表示的是最低構(gòu)造，因為14.00是它的最低位置，它不能向下移動了。 尺寸塊的末端構(gòu)造有兩個在最優(yōu)化階段所應(yīng)用的重要參數(shù)。在兩個尺寸標記中沒有任何重疊的情況下，看測量所有特征是否可行的過程中，這兩個參數(shù)是非常有用的。在形成一種測量末端構(gòu)造中可以看到這兩個參數(shù)是非常有用的。 特點1：在一個尺寸標記處于它的最高（或最低）的位置上，最少有一個尺寸標記處于最高（或最低）的位置。 特點2：只要尺寸塊有末端構(gòu)造，則它就是不正確的。 特點1可以由反證法證明。假如一個尺寸處于它的最高（或最低）構(gòu)造上，并且沒有尺寸標記處于最高（或最低）位置。因此所有它的尺寸標記都不處于它們的最高（或最低）位置，它們能夠同時向上（或向下）移動相同距離直到它們中的任意一個達到了最高（或最低）位置上。因為所有的尺寸標記都是同時移動了相同的距離，尺寸標記就不會重疊，并且因此構(gòu)造結(jié)果仍然是正確的，并且在一個比它初始構(gòu)造更高點（或低點）的位置上。這就違反了初始構(gòu)造處于最高或最低的假設(shè)。 特點2可以直接證明。給定的一個正確構(gòu)造、尺寸塊移動更高（或低）直到一個或更高的尺寸標記達到最高（或最低）位置。因此所有尺寸標記都同時移動相同距離，重疊也不會發(fā)生。另外，因為至少它的一個尺寸標記達到最高（或最低）位置，尺寸塊在沒有不正確構(gòu)造的情況下，尺寸塊不能被移動更高（或更低）。根據(jù)定義2，構(gòu)造的結(jié)果也是最高（或最低）位置。另一方面，尺寸塊是否有末端構(gòu)造就很明顯了，因為末端構(gòu)造定義是正確的所以該尺寸塊就是正確構(gòu)造。 特點1表明尺寸塊的末端構(gòu)造可以由通過觀察塊中的尺寸標記的末端位置的方法獲得。尺寸塊的構(gòu)造可以通過{}來規(guī)定，i=1，2，3，……，n，是特征系列中第i個特征的尺寸標記的位置。這說明了{}是按照他們的垂直上升順序排布的（假如i﹥j，則﹥）。然而，為了避免尺寸標記的重疊，第i個尺寸標記的位置由下式給出： ； n≥i≥2 （2） 式中SIZE是尺寸標記的大小，是該系列中的第一個特征（）尺寸標記的位置。也同樣被用作尺寸塊的參考位置。 如果構(gòu)造是正確的，通過式（1）得出所有尺寸標記必須位于其自己最高位置的下端，如下： ≥ 因此 以上關(guān)系必須通過所有i證明。因此，最高允許值通過給出： （3） 的值通過式（2）給出，一個或更多的都等于。所有其他的都小于它的。沒有更大的值因此構(gòu)造滿足≥，假如正確，這個最終構(gòu)造就是它的最高構(gòu)造。然而，在此構(gòu)造中通過式（2）給出的中的一些值可能小于。因此，每個必須驗證。假如所有的i的≥，則這個最高構(gòu)造成立。假如一些i的，則這個構(gòu)造成立就是不正確的，并且沒有最高位置成立。通過特性2特征系列{}不存在任何正確構(gòu)造。 為了尋找最低位置，對于所有i的≥，最低允許值由下式給出： 對于一個零件來說，它的尺寸塊是多樣化的，假如所有尺寸塊中的兩個末端構(gòu)造能夠成功建立，則在每個沒有任何重疊尺寸塊中放置尺寸標記就是可行的。然而，一個尺寸塊中的一個尺寸標記可能與另一個尺寸塊中的尺寸標記重疊。因此下一步就是測試所有沒有任何重疊的尺寸塊的放置是否可行。下面所講的特點3在解釋這個測試過程就是十分有用的。 特點3：在一個尺寸塊中的兩個末端構(gòu)造之間通常建立一個正確的構(gòu)造。 在特點3中很容易發(fā)現(xiàn)，從最低位置開始，在尺寸塊中的尺寸標記能夠同時移動相同的距離，以使它的尺寸標記能夠達到它的最高位置。根據(jù)定義1，此位置的構(gòu)造是正確的，當(dāng)它的一個尺寸標記達到了最高位置，就達到了尺寸塊的最高位置。 為了測試放置沒有重疊的所有尺寸塊的可行性，一個特征系列的所有特征首先參考按照它們特征的上升順序挑選。然后所有特征系列根據(jù)它們的第一個特征的位置按照上升順序排列。第一個尺寸塊被放置在它們最低構(gòu)造的位置。對于第二個特征系列，假如比的頂端高，尺寸塊也被放置在它們最低構(gòu)造的位置，否則在和的范圍內(nèi)放在的頂端。根據(jù)特點3，對于后者也是正確構(gòu)造。否則，作為正確的構(gòu)造在沒有與尺寸塊重疊的情況下也不能夠建立。對于下個尺寸塊來說是一個重復(fù)的過程，直到所有正確的尺寸塊都已建立，否則無論對于一個尺寸塊正確的構(gòu)造不能建立的話都將中止。 4.2最優(yōu)化階段 在準備階段之后，每個尺寸塊的邊緣構(gòu)造已經(jīng)建立，并且使在一個尺寸塊和兩個相鄰尺寸塊確定避免。使用一種動態(tài)工程方法來確定每個尺寸塊最優(yōu)構(gòu)造。像上面假設(shè)一樣，根據(jù)尺寸塊的參考位置用上升順序來挑選尺寸塊。確定尺寸塊構(gòu)造的過程被分為步，i=1，2，3，……，n，在第i步確定了尺寸塊的構(gòu)造。在每個步，的每個可能構(gòu)造與其狀態(tài)相聯(lián)系。換句話說，狀態(tài)與第i個尺寸塊的第j個構(gòu)造相聯(lián)系。選取第步的狀態(tài)，的費用通過一個總的費用函數(shù)來反映，它是通過下式給出： 是費用函數(shù)用來反映：（i）在獨立的狀態(tài)和的情況在兩尺寸塊和的相互作用；（ii）尺寸塊從它缺省構(gòu)造中的偏差范圍；（iii）兩尺寸塊和之間的重疊；（iv）尺寸塊與被稱為禁區(qū)的一系列區(qū)域之間的重疊。禁區(qū)由用戶規(guī)定，并且禁區(qū)通常為由用戶放置的其他尺寸標記的區(qū)域，因此不允許把尺寸標記放置在遠的地方。最優(yōu)的解決方法從中獲得。這個步驟系列=1，2，3，……，n使最優(yōu)化得以提高，同時這一系列構(gòu)造使尺寸標記的放置位置得到最優(yōu)化。 4.3狀態(tài)解決方法 通過特點3，一個尺寸塊的兩個邊緣構(gòu)造之間的構(gòu)造是正確的構(gòu)造，因此每一步都有無數(shù)種選擇。為了使動態(tài)工程方法起作用，必須使用離散化使無數(shù)的狀態(tài)為每個步。離散化最簡單的方法是從 和的位置中提取一個合適的數(shù)字，這種方法對利用計算機資源來說不是一個有效的方法。這是因為尺寸塊的范圍（由-給定）可以很大限度的改變，從式1中可以明顯的看出尺寸塊給出了一個很大的范圍，而很少值的給了一個很小的范圍。在有合適數(shù)量的狀態(tài)下，那些有較大范圍的尺寸塊會得到近似的解決方法。另一方面，對于一個有很大范圍的尺寸塊來說，就像大多數(shù)在兩邊緣構(gòu)造的正確構(gòu)造將在它們的尺寸塊重疊。通過一個例子最好的解釋：假如在尺寸塊最高構(gòu)造的頂端高于尺寸塊最高位置構(gòu)造的底端，則尺寸塊在構(gòu)造上必然與的最高端重疊。相似的問題在最低構(gòu)造上。因此，對于每個尺寸塊來說，一個合適的范圍，由和之間的不同來定義： = = 式中和是尺寸塊和中尺寸標記的數(shù)量且是獨立的。使用這種可行的范圍定義，那些在兩相鄰尺寸塊之間經(jīng)常導(dǎo)致重疊的構(gòu)造從可行范圍內(nèi)排除。 4.4費用函數(shù) 一個階段的費用和費用函數(shù)都是矢量和矢量值函數(shù)。一個費用矢量包括按重要性由高到低排列的5個部分，i=1，2，3，……，5。動態(tài)的實施階段要求最小的費用選擇，因此必須在兩矢量間進行對比。兩個費用函數(shù)通過比較它們的組成來比較。比較從認為最重要的第一部分開始。假如兩矢量的第一部分相等，則比較認為是次重要的下一部分。當(dāng)相關(guān)聯(lián)的兩部分不相等時就停止比較。在費用矢量尖的比較是建立在不相等的第一對部件上的。 費用函數(shù)的第一組成部分等式應(yīng)該減少相鄰兩尺寸塊和和尺寸塊和禁區(qū)之間的重疊。由下式給出： 式中如果在和之間沒有重疊=0。如果有重疊，則就是一個很大的值。費用函數(shù)用來減少與狀態(tài)相關(guān)聯(lián)的尺寸塊和禁區(qū)之間的重疊。的所有和m認為是與所有禁區(qū)相重疊的值。在和如果沒有重疊，則=0；如果有重疊，則就是一個很大的值。 接下來的四個等式和，是分配四個可選的亞費用函數(shù)，這些函數(shù)根據(jù)四個不同的基準通過返回值來反映構(gòu)造的。的布局與亞費用函數(shù)由用戶確定。這就給了用戶的可行性來決定準則的重要聯(lián)系。 要求尺寸塊放置在特征的中間部分能夠被測缺省構(gòu)造測量的位置，這就認為是尺寸塊的缺省構(gòu)造。亞費用函數(shù)是用來測量尺寸塊在它的第j個構(gòu)造處偏移的程度。這個偏移值可以從特征的平均位置上來測量，并且是第j個構(gòu)造尺寸標記的平均位置 = 式中為尺寸塊中尺寸標記的數(shù)量，為在第j個構(gòu)造上尺寸標記的位置，為特征的位置。和是所有尺寸標記和特征的數(shù)量。 在只被用作亞費用函數(shù)的時候，不能通過費用函數(shù)來確定。因為在每兩個狀態(tài)下偏移的數(shù)是確定的費用函數(shù)是僅建立在能夠給出相同費用的可選費用函數(shù)上的，可以忽略這個問題?？梢苑峙涞搅硪粋€可以給出一個較低狀態(tài)費用函數(shù)的部件等式。假如尺寸塊在狀態(tài)時不在它的缺省狀態(tài)，取1，否則取0。 可能要求由用戶規(guī)定一個可行范圍的百分比來限制尺寸塊由缺省位置偏移的范圍。亞費用函數(shù)是用來減少過大的偏移。假如>p×︳YFmaxi ?YFmin ︳，則的值為1，否則為0。 當(dāng)兩個相鄰尺寸塊在它們的缺省位置重疊，某個值a，則在此重疊中向上移動距離為，并且向下移動到，距離為，其中+=。其中要求=。這就要求兩尺寸塊移動量相等。亞費用函數(shù)不能達到這個目的，因為這僅是測量移動值不是相鄰尺寸塊移動值的分配。則要求兩相鄰尺寸塊的移動值相等。在兩相鄰尺寸塊移動值的范圍通過下式設(shè)置不同值： 這四個可選的亞費用函數(shù)可以由用戶自由的選擇，安排第二到第五部分來達到目的。在下一部分中，通過例子來說明選擇不同的亞費用函數(shù)達到不同的結(jié)束。 5.實例說明 合適的最優(yōu)化方法已經(jīng)被執(zhí)行，并且通過API插入到UG2系統(tǒng)中。為了說明亞費用函數(shù)的作用，思考圖6中的作用。圖6中有5個尺寸塊，所有尺寸塊都處于它們的缺省構(gòu)造，并且在和之間，和 DOI 10.1007/s00170-004-2374-2 ORIGINAL ARTICLE Int J Adv Manuf Technol (2006) 28: 370–378 C.L. Li · K.M. Yu · Y.H. Lee Automatic datum dimensioning for plastic injection mould design and manufacturing Received: 7 May 2004 / Accepted: 10 August 2004 / Published online: 20 April 2005 ? Springer-Verlag London Limited 2005 Abstract Datum dimensioning (or ordinate dimensioning) tech- nique is very popular in plastic injection mould drawings where the location dimensions of a large number of hole features must be specified in the drawings of the mould plates. Although com- mercial CAD/CAM systems provide semi-automatic tools to as- sist the designer in the dimensioning process, it is still a very tedious process, as the user has to specify the location of each di- mension tag. This paper reports a completely automatic method where optimal placements of the dimension tags can be deter- mined. The method employs dynamic programming technique to optimize the dimension process with respect to several criteria that can be selected by the user. The method has been imple- mented and incorporated into a commercial CAD/CAM system, and examples are given to illustrate the important features of the program. Keywords Automatic dimensioning · Datum dimensioning · Dynamic programming · Optimal dimensioning · Ordinate dimensioning 1 Introduction CAD/CAM systems are now widely used in the plastic injec- tion mould-making industry. Many companies are using a solid modeling system to design the injection mould. They use a CAD system to model not only the core and cavity inserts of the mould (which are the most important components that form the im- pression of the mould), but also all other components in the C.L. Li (a117) · Y. H . L e e Department of Manufacturing Engineering and Engineering Management, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong E-mail: meclli@cityu.edu.hk Tel.: +8-52-27888432 Fax: +8-52-27888423 K.M. Yu Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University entire mould assembly. With the advance in Internet technology and the recent development of Internet-enabled CAD, the de- sign information of the injection mould can be communicated electronically between the product engineer (who designs the plastic part) and the tooling engineer (who designs the injection mould), even though they may be located in different geographic regions of the world. While flow of design information between product design and tooling design are communicated effectively through an electronic means, the communication of manufac- turing information to the shop floor is done by both electronic and traditional techniques. Computer Numerical Control (CNC) machining toolpath or inspection instructions can be generated directly from the CAD/CAM system and downloaded through a network to the CNC controller for the machining or inspection operations. However, set-up instructions for a particular machin- ing job may be specified in an engineering drawing. Moreover, not all machining tasks are done using CNC machine tools. Some traditional machining processes, such as drilling and grinding, are done using conventional machine tools because of cost con- sideration. Conventional engineering drawings are thus still play- ing an important role in communicating engineering information to the shop floor. The orthographic projections in engineering drawings can be generated automatically from the CAD model of the parts. Automatic tools for dimensioning of the parts are also provided by many commercial CAD systems. However, as pointed out by Chen et al. [1], those automatic dimensioning tools are not able to generate dimensions according to the draw- ing standards and engineering practices adopted in the shop floor. In the specific application of injection mould design, datum dimensioning (or ordinate dimensioning) of hole features are used extensively. Figure 1 shows a typical detail drawing that can be found on the shop floor of a mould making company. Shown in the figure are the hole features and datum dimensions which are used to specify the locations of the holes. It can be seen that the dimensions are very crowded and it is a tedious task to manually adjust the placement of all the datum dimensions. The quality of the final fully-dimensioned drawing thus depends very much on the experience of the draftsman who produces the draw- ing. The purpose of this research is to develop a tool that can 371 Fig. 1. Use of datum dimensioning in a drawing of a plastic injection mould part generate the datum dimensions automatically from a given part of the injection mould. The resulting dimensions must satisfy two obvious requirements: first, that no two dimension tags may overlap; and second, that a dimension tag be placed as close as possible to the feature being dimensioned. The key issue in this research is to develop a method that can optimize the placement of the datum dimensions. 2 Related work While dimensioning and tolerancing are two closely related pro- cesses in specifying the size and location information of the features in a mechanical part or an assembly, most of the past research work has focused on tolerancing. The major research issues in tolerancing are representation, analysis and synthesis. Tolerancing representation is concerned with the incorporation of tolerance information into a product modeling scheme. Exam- ples include the solid offset approach developed by Requicha [2], the feasibility space approach proposed by Turner [3], and the TTRS by Desrochers and Clement [4]. More detailed review can be found in Roy et al. [5] and Yu et al. [6]. Tolerance analy- sis aims to determine the combined effect of part tolerances on the assembly tolerance. It can be used to verify the functional- ity of a design given known or assumed variations of individual part dimensions. Examples of technique in tolerance analysis in- clude Monte Carlo simulation [7] and the direct linearization method [8]. The main objective of tolerance synthesis or tol- erance allocation is to allocate part tolerances based on given functional requirements of the assembly. Recently, Islam [9] re- ported a concurrent engineering approach to address this prob- lem. Based on a systemic analysis of the functional requirements from different customer requirements and the technical require- ments from engineering considerations, a methodology for ex- tracting dimensional requirements is developed. A software pro- totype FDT [10] is also developed for supporting the implemen- tation of the methodology. FDT provides tools for representing the functional requirements, dimensions, tolerances and process capability into a functional requirement/dimensions matrix. The functional equations captured in the matrix are then separated into groups, and each group is then solved using a solution strat- egy specific to the functional requirement and the tolerancing problem involved. More detailed review in tolerance analysis and synthesis can be found in Roy et al. [5], Ngoi and Ong [11] and Hong and Chang [12]. Several methods have been developed for generating dimen- sions automatically from the CAD model of a part. Yuen et al. [13] reported an early attempt in automatic dimensioning of parts represented in Constructive Solid Geometry (CSG) solid modeling technique. Points from planar faces and axes of cylin- ders are extracted from the solid model. The coordinates of the points are arranged in a tree structure to generate linear dimen- sions in the three principal directions. A simple technique for diametric and radial dimensions was also reported. Other early works in automatic dimensioning have been summarized by Yu et al. [6]. Recently, Chen et al. [1, 14] reported a more in-depth study of automatic dimensioning. Their method analyzed di- mension redundancy, determined dimensioning schemes that are specific to feature patterns, selected appropriate views for spec- ifying the dimension, and determined the appropriate location of the dimension using an expert system approach [15]. The ex- pert system analyses the geometry and topology of the feature to be dimensioned, and determined a position for placing the dimension based on a set of rules that is relevant to the cur- rent dimensioning feature. With the placement of one dimension, a forbidden region is constructed so that all subsequent dimen- sions will not be placed in this region. This avoids overlap or intersection between two dimensions. 372 A limitation in the existing approach for the placement of the dimension is due to the sequential nature of the method. For example, in Chen’s [1, 14] method the features to be dimen- sioned are prioritized, and the positions of the dimensions are determined one after another. The approach is not appropriate for determining the placement of datum dimensions, especially when the dimensions are very crowded, as in the case of injection mould plates. This is because the placement of one datum di- mension may have an effect on the placement of another dimen- sion that may be located far away from the current dimension. This paper reports our work in solving the placement problem in datum dimensioning. The major contribution of our work is the development of a new method that determines the optimal placement of each datum dimension. Using the dynamic pro- gramming approach to optimization, this new method overcomes the limitation of the sequential approach used in the existing method. 3 Basic characteristic of datum dimensioning In datum dimensioning, the location of a feature is specified by the horizontal and vertical distances from the reference lo- cation of the feature and a reference datum. The default form of datum dimension is shown in Fig. 2a. When the vertical dis- tance between two features to be dimensioned is less than the dimension tag size (i.e. the sum of the dimension text height and the minimum spacing between adjacent dimension texts), Fig. 2. Basic characteristics of datum dimensioning the alternative forms shown in Fig. 2b are required. 1 The di- mension tags are shifted upward or downward from the default location to prevent overlap. As shown in Fig. 2c, the shifting of the dimension tag is achieved by breaking the single exten- sion line of the dimension into three segments: two horizontal segments which are connected by one inclined segment. The ex- tent to which a dimension tag can be shifted is governed by three parameters: (i) the dogleg angle α, which is the angle be- tween the inclined segment and the horizontal segments of the dimension line; (ii) the margin distance m between the dimen- sion text and the part boundary; and (iii) the location (x f i , y f i ) of the feature f i . The two extreme positions (i.e. the uppermost pos- ition y max i and lowermost position y min i ) of the dimension tag are given by: y max i = y f i +(x f i +m) tan α y min i = y f i ?(x f i +m) tan α (1) 4 Automatic datum dimension The objective of the automatic datum dimensioning system is to find an optimal position for each datum dimension. The process consists of two phases of operation: the preparation phase and the optimization phase. In the preparation phase, major param- eters that facilitate the optimization process will be established. Feasibility for placing the dimensions for all the features using the given dogleg angle, margin offset and dimension tag size will also be tested. In the optimization phase, a dynamic pro- gramming approach is used. The dimension tag locations can be optimized with respect to different sets of criteria, including the minimization of the shift of every dimension from their default locations, or maximization of the use of the default form as much as possible. 4.1 The preparation phase The features to be dimensioned are first grouped into one or more feature sets. For each feature in a feature set, there exist at least one other feature in the set such that the vertical dis- tance between them is less than the dimension tag size. In other words, the features in a feature set cannot be dimensioned using the default form exclusively without overlap between adjacent dimension tags. Instead, at most one feature can use the de- fault form while all others require the use of the alternative form. The set of dimension tags associated with a feature set is called a dimension block. The configuration of a dimension block refers to the forms and locations of each datum dimen- sion within the dimension block. For each position of a dimen- sion block, its configuration is uniquely defined. Figure 3 shows two feature sets and their dimension blocks at two different configurations. 1 To simplify the explanation of the technique, only vertical dimensions placed on the left hand side of the part are discussed. The method developed is general and can be applied to the other sides of the part. 373 Fig. 3. Feature sets and different con- figurations of dimension blocks Definition 1: Validity of a configuration. A configuration of a di- mension block is valid if there is no overlap between any dimen- sion tags in the dimension block, and each dimension tag lies within its extreme positions. The configurations of the dimension blocks shown in Fig. 3b are valid. Two examples of invalid configuration are shown in Fig. 4. The configuration shown in Fig. 4a is invalid because two of the dimension tags overlap. For the configuration shown in Fig. 4b, the extension line of the dimension tag 14.00 is at its lowermost position, while the required position for the dimen- sion tag is beyond this lowermost position. Fig. 4. Invalid configurations of a dimension block Fig. 5. Dimension block at extreme configurations Definition 2: Extreme configurations. There are two extreme configurations: the uppermost and lowermost configurations. A dimension block is at its uppermost (lowermost) configuration if the dimension block is valid and is at a position such that any other higher (lower) position results in an invalid configuration. The extreme configurations of a dimension block d i are denoted by Y max i and Y min i . Figure 5a shows a dimension block at its uppermost config- uration. It cannot move further upward because the dimension tag 29.5 is at its highest position. Figure 5b shows a dimen- sion block at its lowermost configuration. It cannot move fur- 374 ther downward because the dimension tag 14.00 is at its lowest position. The extreme configurations of a dimension block are the two important parameters that will be used by the optimization pro- cess. They are also useful in testing whether it is feasible to dimension all the features without any overlap between the di- mension tags. It is observed that two properties are useful in developing a method to determine the extreme configurations. Property 1:. For a dimension block at its uppermost (lowermost) configuration, at least one of its dimension tags is at its upper- most (lowermost) position. Property 2:. A dimension block has a valid configuration if and only if it has extreme configurations. Property 1 can be proved by contradiction. Assume that a di- mension block is at its uppermost (lowermost) configuration, and none of its dimension tags are at their uppermost (lowermost) positions. Since all the dimension tags are not at their uppermost (lowermost) positions, they can all be moved upwards (down- wards) simultaneously by the same amount until any one of them reaches its uppermost (lowermost) position. As all dimension tags are moved simultaneously by the same amount, the dimen- sion tags do not overlap, and thus the resulting configuration is still valid and at a higher (lower) position than its original configuration. This violates the assumption that the original con- figuration is the uppermost (lowermost) configuration. Property 2 can be verified directly. Given a valid configura- tion, the dimension block is moved upward (downward) until one or more of its dimension tags reach its uppermost (lowermost) position. Since all the dimension tags are moved simultaneously by the same amount, overlap does not occur. Moreover, the di- mension block cannot be moved upwards (downwards) any fur- ther without invalidating the configuration because at least one of its dimension tags is at its uppermost (lowermost) position. According to Definition 2, the resulting configuration is thus the uppermost (lowermost) configuration. On the other hand, it is ob- vious that if a dimension block has extreme configurations, then it has a valid configuration because the extreme configurations are, by definition, valid. Property 1 indicates that the extreme configurations of a di- mension block can be obtained by investigating the extreme pos- itions of the dimension tags in the block. The configuration of a dimension block can be specified by {y i }, i = 1, 2,...,n,where y i is the location of the dimension tag of the ith feature in the feature set { f i }.Thisassumesthat{ f i } are arranged in ascend- ing order by their vertical positions (i.e. y f i > y f j if i > j). Then, to avoid overlap between dimension tags, the location of the ith dimension tag is given by: y i = (i ?1)×SIZE + y 1 ; n ≥ i ≥ 2(2) where SIZE is the dimension tag size and y 1 is the location of the dimension tag for the first feature ( f 1 )oftheset.y 1 is also used as the reference location of the dimension block. For a configuration to be valid, all dimension tags must lie below its own uppermost position given by Eq. 1. That is: y max i ≥ y i and thus y max i ≥ (i ?1)×SIZE + y 1 The above relationship must be satisfied by all i. Therefore, the highest allowable value for y 1 is given by: Min i {y max i ?(i ?1)×SIZE} (3) with the y 1 value given by Eq. 2, and one or more y i equal to y max i . All other y i are less than its y max i . Since no other larger value of y 1 results in a configuration that satisfies y max i > y i ,the resulting configuration, if valid, is the uppermost configuration. However, it is possible that at this configuration some of the y i given by Eq. 2 is less than y min i . Therefore, a check is performed for each y i .Ify i ≥ y min i for all i, then the uppermost configura- tion is found. If y i i (Y max j ?SIZE×n i )) YF min i = Max(Y min i , Max i> j≥1 (Y min j +SIZE×n j )) where n i and n j are the number of dimension tags in dimension blocks d i and d j , respectively. Using this definition of the feas- ible range, those configurations that always cause overlap with adjacent dimension blocks are excluded from the feasible range. A fixed resolution, say 0.5 mm, is specified and the number of states for a given stage is obtained by dividing the feasible range by the given resolution. 4.4 Cost functions The overall cost of a stage and the cost function C i (t i, j , t i?1,k ) are vector and vector-valued functions, respectively. A cost vec- tor consists of five components c i , i = 1,...,5arrangedinde- scending order of importance. That is, c i is considered more important than c j if i p× vextendsingle vextendsingle YF max i ?YF min i vextendsingle vextendsingle , and is set to zero otherwise. When two adjacent dimension blocks at their default loca- tions overlap for a certain amount a, the overlap can be removed by shifting d i upwards by a i , and shifting d i?1 downwards by a i?1 , such that a i +a i?1 = a. It may be desirable that a i = a i?1 . That is, the required total shift is being shared equally between two dimension blocks. The sub-cost function D V (t i, j ) is not able to achieve this purpose because it only measures the total shift amount and not the distribution of the amount between adjacent dimension blocks. D E (t i, j ) is devised to equalize the amount of shift between adjacent dimension blocks. It is set to the differ- ence between the extent of deviations of the adjacent dimension blocks. D E (t i, j , t i?1,k ) =|D V (t i, j )? D V (t i?1,k )|. The four optional sub-