【溫馨提示】壓縮包內(nèi)含CAD圖有下方大圖片預(yù)覽,下拉即可直觀呈現(xiàn)眼前查看、盡收眼底縱觀。打包內(nèi)容里dwg后綴的文件為CAD圖,可編輯,無水印,高清圖,壓縮包內(nèi)文檔可直接點開預(yù)覽,需要原稿請自助充值下載,所見才能所得,請見壓縮包內(nèi)的文件及下方預(yù)覽,請細心查看有疑問可以咨詢QQ:11970985或197216396
畢業(yè)設(shè)計(論文)外文資料翻譯
系 部: 機械工程系
專 業(yè): 機械工程及自動化
姓 名:
學(xué) 號:
外文出處: Journal of Engieering Materials
and Technology
附 件: 1.外文資料翻譯譯文;2.外文原文。
指導(dǎo)教師評語:
譯文基本能表達原文思想,語句較流暢,條理較清晰,專業(yè)用語翻譯基本準確,基本符合中文習(xí)慣,整體翻譯質(zhì)量一般。
簽名:
年 月 日
附件1:外文資料翻譯譯文
加強使用粘結(jié)劑力軌跡和神經(jīng)網(wǎng)絡(luò)控制
獲得一致的最小回彈
曹健 Brad Kinsey
機械工程系,西北大學(xué),埃文斯頓,IL 60208
Sara A. Solla
物理和天文學(xué)系,西北大學(xué),埃文斯頓,IL 60208;
生理學(xué)系,西北大學(xué)醫(yī)學(xué)院,芝加哥,IL 60611
制造鈑金屬片面臨的最大挑戰(zhàn)之一是要取得一致的回彈量?;貜?,當工件加工撤掉后,彈性材料恢復(fù),主要的原因是最后一部分的幾何變化的不一致。由于非線性效應(yīng),相互作用過程和材料參數(shù)的關(guān)系,要獲得一致可取的回彈量是極其困難。在本文中,回彈量在模擬通道的形成過程證明了神經(jīng)網(wǎng)絡(luò)系統(tǒng)的獨特的能力同時加上加強粘結(jié)劑力軌跡控制回彈角和最大主應(yīng)變量起的作用。當面臨著很大差異甚至在材料性能,板度,摩擦條件下,我們的控制系統(tǒng)將會制作出零件最終的形狀。
導(dǎo)言
在當今競爭激烈的制造業(yè)中零件獲得一致準確的尺寸是至關(guān)重要的。不一致的零件尺寸會延緩新產(chǎn)品的開發(fā),增加轉(zhuǎn)換時間,使后續(xù)開發(fā)生產(chǎn)困難,還需要額外的保證工序質(zhì)量,同時還降低客戶對終端產(chǎn)品的滿意度和信任度。在金屬板料成形過程中,回彈方面的材料是保證零件最終精確的一個主要因素?;貜検窃谄錆M載的條件下幾何的差額部分,例如,符合工件的幾何形狀,當局部已卸下,自由的狀態(tài),幾何之間的差額。對于一個復(fù)雜的三維零件,不必要的扭曲是另一種形式的回彈。薄板厚度方向中壓力分布不一致和沖壓加載放寬和卸載過程中,其結(jié)果是產(chǎn)生回彈。影響回彈的因素包括變化過程和材料參數(shù),如摩擦條件下,工件的幾何形狀,材料性能,板的厚度,以及模具溫度。由于在制造過程中控制所有這些變數(shù)是幾乎不可能的,反過來證明回彈是不能隨時控制?;貜検且粋€高度非線性效應(yīng),這個新增加的困難是事實,因此,模擬和糾正的方法是復(fù)雜的。最近幾年出現(xiàn)了許多研究有關(guān)回彈,例如,汽車工程師學(xué)會,NUMIFORM和NUMISHEET會議。
通過修改形成過程可減少回彈。一些研究人員提出了使用加強粘結(jié)劑力軌跡以實現(xiàn)這一目標(Ayres,1984;Hishida and Wagoner, 1993; Sunseri et al., 1996)。進一步加強粘結(jié)劑力軌跡是一個瞬時跳躍從低粘結(jié)力(LBF)到高粘合劑力(HBF),在指定的百分比總額的沖壓位移(PD%)(參閱圖1)。Sunseri et al. (1996),調(diào)查鋁通道形成過程的回彈,圖2所示。他們的工作是通過模擬實驗明確形成過程和材料參數(shù)用處。在生產(chǎn)環(huán)境中,可是,由于不同的進程回彈量將偏離所期望的水平。因此,控制系統(tǒng),可變化的工藝參數(shù)要求
近年來,許多研究小組調(diào)查了利用人工神經(jīng)網(wǎng)絡(luò)系統(tǒng)來控制金屬板料成形過程。由于非線性效應(yīng)和相互作用的過程參數(shù)金屬成型是an ideal candidate,神經(jīng)網(wǎng)絡(luò)系統(tǒng)控制。(Cho et al. (1997)采用了神經(jīng)網(wǎng)絡(luò)預(yù)測的力量在冷軋和,(Di and Thomson(1997)預(yù)測起皺限制正方米板材下角的緊拉應(yīng)力。除此之外,(Elkins, 1994 and Yang et al., 1992), Forcellese et al. (1997)采用了神經(jīng)網(wǎng)絡(luò)系統(tǒng)控制60度鋁V型沖壓空氣彎曲回彈的進程。他們的系統(tǒng)利用實驗獲得的例子組成的5個參數(shù)從力沖壓軌跡,離線測量板厚,神經(jīng)網(wǎng)絡(luò)目標彎曲角度的投入和沖壓的移位的輸出。在另一項研究項目,Ruffini and Cao(1998)建議采用神經(jīng)網(wǎng)絡(luò)控制回彈角的渠道的形成過程與軌跡沖壓力作為唯一的來源,確定工藝變化和調(diào)整的HBF,Sunseri et al. (1996)。初步結(jié)果表明這種方法是很有希望的。
本文中,鋁渠道的回彈是通過加強控制粘結(jié)劑力軌跡和神經(jīng)網(wǎng)絡(luò)控制。神經(jīng)網(wǎng)絡(luò)確定了HBF和PD%的粘結(jié)劑力軌跡。沖床力軌跡被確定為關(guān)鍵參數(shù)的變化,反映在材料性能,板的厚度,還有摩擦系數(shù)。因此,四個多項式系數(shù)從曲線擬合的沖壓力軌跡被用來作為投入的神經(jīng)網(wǎng)絡(luò)系統(tǒng)。圖3顯示流程圖的控制系統(tǒng)提出這一點應(yīng)用。盡管材料性能有著很大差異,板厚度(T),和摩擦系數(shù)(m),回彈角(θ)之間保持了0.2和0.6度,(e)項限于8%和10%。最后,由Sunseri et al.(1996)提出一個比較閉環(huán)控制方法,是為了顯示我們控制方法的好處。雖然模擬結(jié)果在這里只有數(shù)值,然而控制系統(tǒng)將在未來改進需求,以驗證實際執(zhí)行情況。
通道的形成過程
調(diào)查回彈在一個通道的形成過程是一個簡單的幾何學(xué)。因此,由受雇這里得Sunseri et al.(1996)提出使用鋁通道形成過程(參閱圖.2 ) 。
第一,連續(xù)的粘結(jié)劑力(CBF)對影響回彈進行了評價。由于CBF增加,我們的模擬在所有其他進程和材料參數(shù)進行了不斷實驗,回彈角(u)降低,圖形顯示圖4,物理圖5,然而,增加粘結(jié)劑力同時還將造成材料的最大的拉緊力增加,實線圖4,鋁通道的突出(Graf and Hosford, 1993)。通過利用粘結(jié)劑加強力軌跡,適度的增加側(cè)壁得到應(yīng)變力水平,同時減少回彈的過程中體現(xiàn)在表中的圖1。
當面臨著偏差的摩擦系數(shù),只是為了產(chǎn)生一個健全的進程,Sunseri et al.(1996)實施閉環(huán)變量粘結(jié)劑力控制后續(xù)零件的沖壓力軌跡從局部加強粘結(jié)劑的情況制定工藝條件。當摩擦系數(shù)在不同的用途系數(shù)為0.1,甚至是到0.25,這種控制方法能產(chǎn)生回彈的水平一致。然而,這種方法是否可以承受其他參數(shù)的變化,如材料性能和板料厚還沒有確定的情況下。
商業(yè)有限元分析套件(ABAQUS,1997年)被我們用于形成過程中的數(shù)值模擬通道,因為這個問題是接近平面變形應(yīng)力條件,并對稱,只有十六分之一的寬度和一半以上的長度在整個空白(220毫米346毫米)為藍本。粘結(jié)劑,模具和沖壓是仿照作為三個獨立的硬質(zhì)表面。每個面是仿照采用四節(jié)點界面元素(ABAQUS類型IRS4)和庫侖摩擦法是假設(shè)。我們的空白網(wǎng)孔40四節(jié)點分配不均,減少了一體化殼元素(ABAQUS類型S4R)更致密的沖壓和彎道半徑粘結(jié)劑集中接觸在空白的網(wǎng)孔處。邊界條件指定創(chuàng)建一個平面應(yīng)變條件。該材料是各向同性的模板,彈塑性后,馮米塞斯屈服準則和各向同性應(yīng)變硬化。彈性性能的楊氏模量中,E,,70千兆和泊松比,n,0.3。塑料薄板材料的參數(shù)是模板,可使用一些教學(xué)關(guān)系(s5Ken))。我們的名義材料,指材料1,有材料強度系數(shù),K,528兆帕和應(yīng)變硬化指數(shù),n,0.265有限元模型
擬議控制系統(tǒng)
在一個通道的形成過程,回彈一般是極為敏感的材料的變化和成形參數(shù)。在這項工作中,我們制訂一種方法控制回彈,同時材料通過結(jié)合加強粘結(jié)劑力軌跡和神經(jīng)網(wǎng)絡(luò)控制產(chǎn)生一個可以接受的最大應(yīng)變量。在加強局部粘結(jié)劑軌跡,兩個臨界值需要加以確定,巨大的HBF和總沖壓力的移位的PD%,這兩個參數(shù)的輸出從神經(jīng)網(wǎng)絡(luò)系統(tǒng)。從我們以往的研究經(jīng)驗,在金屬板料成形過程控制(Ruffiniand Cao, 1998;Kinsey and Cao, 1997; and Sunseri et al.,1996) ,沖壓力的軌跡被選定為參數(shù),提供有關(guān)當前進程。因此,多項式系數(shù)從曲線擬合的沖壓力軌跡被用來作為輸入的神經(jīng)網(wǎng)絡(luò)。
我們提出的控制系統(tǒng)流程圖如圖所示圖3,形成過程通常是將著手在連續(xù)使用的正常的粘結(jié)劑力,16千牛頓,深度為10毫米。雖然沖壓的移位會繼續(xù)下去,多項式系數(shù)從曲線擬合的沖壓力軌跡的計算和輸入和輸出的神經(jīng)網(wǎng)絡(luò),該HBF和PD%的粘結(jié)劑力加強軌跡,在指定的沖床的位置將得到及時作出HBF適當?shù)恼{(diào)整,19毫米乘以PD%。
其中系數(shù)10毫米的用處從沖壓力軌跡的計算,選擇有兩個原因。在這個距離,沖壓力軌跡是很精確的,有足夠的數(shù)據(jù)點可以準確擬合發(fā)生的曲線。圖6顯示的效果差異t和m對沖壓武力軌跡從0到10毫米。其次,10毫米距離允許足夠的CPU時間來控制計算方法。假設(shè)沖床速度50毫米/秒,并設(shè)定最低PD
%,57.5%的總沖壓移位,10.925毫米,大約有18.5毫秒可用來計算輸入人工神經(jīng)網(wǎng)絡(luò)和預(yù)測階梯粘結(jié)劑力軌跡的HBF和PD%。曲線擬合和神經(jīng)網(wǎng)絡(luò)的程序?qū)ξ覀兊谋简vII,233兆赫的電腦需要大約10毫秒來計算。在形成過程中如何接近函數(shù)將取決于粘結(jié)劑力速度不同進程的機制。
沖壓力的軌跡被分為三個區(qū)域,區(qū)域A,一個過渡區(qū)域,和B區(qū)(參照圖6)。薄板厚度的用途的改變與區(qū)域A的變化相類似,在這個區(qū)域中二次多項式被用于計算出正確的數(shù)據(jù)。盡管事實是二次多項式只有兩個輸入系數(shù)分別用于本區(qū)域的特點,因為沖壓力軌跡幾乎傳遞給了來源數(shù)據(jù)。在B區(qū),造成不同的傾斜度的主要主要原因是摩擦系數(shù)的變化。網(wǎng)絡(luò)系統(tǒng)線性插值提供了兩個額外的輸入數(shù)據(jù)。因此,一共有四個多項式系數(shù)被用作輸入到我們的神經(jīng)網(wǎng)絡(luò)。
值得注意的是沖壓力軌跡取自數(shù)值模擬,因此,它們是光滑的并由于t和m的變化而多樣性。在實際形成過程中,噪聲數(shù)據(jù)采集設(shè)備將產(chǎn)生的差額用這些曲線表達出來。通過使用多項式系數(shù)來表達曲線擬合的沖壓力軌跡,輸入數(shù)據(jù)到神經(jīng)網(wǎng)絡(luò)時的本質(zhì)是要過濾掉實踐時考慮的因素。
為了實施這控制系統(tǒng)在實際成形技術(shù),應(yīng)用的數(shù)據(jù)是需要我們整理出來。該資料可從嘗試通過改變沖壓模具工藝參數(shù)在實際生產(chǎn)中的得到。舉例來說,考慮到一批一批的材料變化,材料可以得到和建立數(shù)據(jù),它體現(xiàn)了金屬板厚度和材料性質(zhì)變化的規(guī)律。此外,潤滑狀態(tài)可能是多種多樣的。利用各種組合的這些工藝參數(shù),該HBF和PD%的價值,產(chǎn)生理想的回彈量,可確定。此外,數(shù)值模擬,校準實驗結(jié)果,以確保準確性,可以用來迅速增加的訓(xùn)練數(shù)據(jù)。一旦神經(jīng)網(wǎng)絡(luò)是訓(xùn)練有素的整個一系列潛在的工藝參數(shù)值,實際值的材料性能,板厚度,摩擦狀態(tài)并不需要,因為投入神經(jīng)網(wǎng)絡(luò)的多項式系數(shù)從沖壓力軌跡曲線擬合。
任何實施增加粘結(jié)劑力軌跡,可以減少回彈。然而,簡單的加強粘結(jié)劑力軌跡,只需要兩個輸入?yún)?shù),HBF和PD%。此外,現(xiàn)有在工業(yè)上的擠壓有能力產(chǎn)生局部加強粘結(jié)劑力軌跡。因此,局部加強粘結(jié)劑力軌跡是一個理想的選擇。
神經(jīng)網(wǎng)絡(luò)
人工神經(jīng)網(wǎng)絡(luò)已研究多年,希望能模仿人類大腦的能力,解決那些模糊的需要大量處理的問題。人類大腦為了實現(xiàn)這些數(shù)據(jù)的處理,利用大規(guī)模并行處理能力,數(shù)以百萬計的神經(jīng)元的共同努力,來解決這些復(fù)雜的問題。同樣,人工神經(jīng)網(wǎng)絡(luò)模型包含許多計算單元,稱謂的“神經(jīng)元”,以符合其生物對位,并聯(lián)運行,并連接與變權(quán)重組合預(yù)測方法。這些重量都在適應(yīng)訓(xùn)練過程,最常見的通過反傳算法(Rumelhart and McClelland, 1986),提出了神經(jīng)網(wǎng)絡(luò),例如輸入產(chǎn)出對的關(guān)系網(wǎng)絡(luò)正在試圖了解。這個目標是神經(jīng)網(wǎng)絡(luò)推論,或引用,模式給出了輸入輸出的例子。神經(jīng)網(wǎng)絡(luò)進一步細節(jié),一般被發(fā)現(xiàn)(Widrow and Lehr (1990)。對于我們的特殊應(yīng)用,神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu)被確定為四個輸入?yún)?shù),5個隱藏的神經(jīng)元,和兩個輸出。一個S形是激活功能是用于隱藏神經(jīng)元,同時利用線性產(chǎn)出。Kinsey(1998)提出了,如何優(yōu)化結(jié)構(gòu)更詳細地結(jié)論。
最初確定利用神經(jīng)網(wǎng)絡(luò)來控制和減少回彈的可行性,神經(jīng)網(wǎng)絡(luò)的能力是處理大量不同的t和m進行研究。板材厚度值從0.8到1.4毫米(從0.1毫米遞增至0.8或1.2毫米),摩擦系數(shù)的水平從0.04至0.20(從0.01遞增到0.04和0.12之間,及以后遞增至0.02)進行了考慮。這些變化在t和m值明顯大于那些在實際大規(guī)模生產(chǎn)形成過程會被視作的值,但用在這里的示范目的,以顯示神經(jīng)網(wǎng)絡(luò)控制甚至相當大的變化的系數(shù)的能力。實驗數(shù)據(jù)是通過試驗和錯誤模擬104組合這兩個工藝參數(shù),以確定HBF和PD%的值,這些值產(chǎn)生了回彈的角度,θ,范圍在0.4至0.5度,最大應(yīng)力,e,范圍在8%至10%。這個極其狹窄的范圍內(nèi),回彈范圍有一定的接近真實的值預(yù)測HBF和PD%的值,實驗網(wǎng)絡(luò)可能不會導(dǎo)致回彈數(shù)值在相同的窄幅波動。在指定的范圍給出的HBF和PD%的組合不一定是唯一一個能夠提供u和E的值。但是,通過有選擇u和e的小范圍變化,我們保證網(wǎng)絡(luò)將接受一個可能的HBF和PD%的值的狹窄的窗口實驗數(shù)據(jù)。這些104模擬運行提供了輸入輸出對,實驗有四個數(shù)據(jù)曲線擬合多項式系數(shù)從沖力軌跡作為輸入,HBF和PD%占所期望的輸出。
結(jié)果
一旦神經(jīng)網(wǎng)絡(luò)有更好的實訓(xùn),在網(wǎng)絡(luò)預(yù)測中4t和m組合實訓(xùn)集排除以及另外4t和m的組合不包括在以前的一系列t和m值分別為“fedforward”,HBF和PD%值為加強局部軌跡粘結(jié)劑,以下所述的程序圖3,先前章節(jié)。由此產(chǎn)生的回彈角,u和最大應(yīng)力,e,從這一進程中,然后計算實驗數(shù)據(jù)。表1顯示了出色的成績,得到了這8數(shù)據(jù)結(jié)果。所有的回彈角和最大應(yīng)力值分別在0.3至0.6度和8%至10%范圍之間。
神經(jīng)網(wǎng)絡(luò)的能力,提供HBF和PD%的值使用,面臨著t和m很大的差異顯示在實際應(yīng)用神經(jīng)網(wǎng)絡(luò)的潛力。然而,過程中還有其他參數(shù)的變化,對于最后部分的形狀產(chǎn)生類似的效果。材料特性,例如,已被證明造成嚴重的三維變化的金屬薄板沖壓件(Kinsey and Cao, 1997)。當原先的實訓(xùn)網(wǎng)絡(luò)用來預(yù)測不同的材料特性HBF和PD%值,會產(chǎn)生不符合的回彈水平和最大應(yīng)力。這并不奇怪,因為網(wǎng)絡(luò)的實訓(xùn)并未適應(yīng)差異的材料性質(zhì)。因此,更多的實驗結(jié)果數(shù)據(jù)包含材料性質(zhì)是必要的。偏差的真實應(yīng)力應(yīng)變曲線的名義材料,材料1,創(chuàng)造了不同的強度系數(shù),K,由+10%(材料2和3)和+20%(材料第6和第7)和應(yīng)變硬化部分,n,約+16%(材料4和5)。7個組合(t=0.9/u=0.04,t=1.1/u=0.06,=1.4t/=0.08,t=1.0/u=0.10,t=0.8/m=0.12t=1.0/u=0.16,與t=1.2/u=0.20)被選為該實驗的例子與新材料的創(chuàng)建。再次通過反復(fù)試驗,某一假定材料的36種實驗范例,t,與u組合產(chǎn)生了相同的變化范圍,比如以前使用的θ和ε,分別為0.4到0.5度和8%至10%。其中6個新型材料,每t和m的組合,將被用來作為檢查而被排除在實驗外,看看是否符合神經(jīng)網(wǎng)絡(luò)預(yù)測準確的HBF和PD%值。這個網(wǎng)絡(luò)使用以前同樣的神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu)。,經(jīng)過這些額外的實驗與變化的例子,材料性能又增加了實訓(xùn)集。
表2顯示的結(jié)果的,當這結(jié)果從被排除在實訓(xùn)集的六個組合,和4個新材料厚度,t,和摩擦系數(shù),u,在新實驗網(wǎng)絡(luò)組合實訓(xùn)。另外,神經(jīng)網(wǎng)絡(luò)能夠提供可接受的的HBF和PD%值,即產(chǎn)生可接受θ0.2至0.6度,e,8%至10%。然而,作為獲得的t和θ的結(jié)果,這些模擬產(chǎn)出的數(shù)據(jù)并不總是在0.3至0.6度這個狹窄的回彈角范圍內(nèi)。這表明,神經(jīng)網(wǎng)絡(luò)在材料中具有很多復(fù)雜的變化。如前所述,范圍狹窄對u在實驗中故意造成被允許的合理差異的前饋過程。此外,更多的培訓(xùn),集或一個更聰明的神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu),如給予“提示”網(wǎng)絡(luò)有關(guān)的材料特性,能提高能力的神經(jīng)網(wǎng)絡(luò)處理變化的材料特性。
閉環(huán)控制的比較結(jié)果
Sunseri et al. (1996)控制系統(tǒng)提出的,利用閉環(huán)控制的粘結(jié)劑局部按照沖壓力軌跡,在同一通道的形成過程以控制回彈。關(guān)于摩擦系數(shù)的可能是最重要的工藝參數(shù),不同的摩擦系數(shù)在控制系統(tǒng)在進行測試,與數(shù)據(jù)獲得。但是,他們的工作是調(diào)查不同的材料特性和板厚。因此,進一步閉環(huán)控制模擬與這些變化進行了這里,以形成一個比較神經(jīng)網(wǎng)絡(luò)控制系統(tǒng)。
首先,從我們的軌跡案名義的沖壓局部力,已創(chuàng)建的材料1,1.0mm的t,和0.10的m,。加比例積分(PI)的控制器,用于調(diào)整粘結(jié)劑力,使下面的沖壓力軌跡,Sunseriet al. (1996)。最大位移之間的沖壓
調(diào)整的粘結(jié)劑力為0.02毫米。表3所列的比例增加kp,以及整體調(diào)整ki,所使用的控制器的3個事件進行數(shù)據(jù)調(diào)查。圖7顯示如何更好的追蹤沖壓力軌跡沖壓軌跡這三個事件(材料4ki50.8)。圖8顯示了如何在不同的粘合劑力的沖壓的移位,使下面的沖壓力軌跡(材料4ki50.8)。請注意,粘結(jié)劑力軌跡趨于平穩(wěn)結(jié)束時的形成過程類似于粘結(jié)劑加強局部的軌道。
從表3顯示從這些閉環(huán)控制實驗得到的θ和e的值同從神經(jīng)網(wǎng)絡(luò)控制系統(tǒng)的結(jié)果聯(lián)系。此表清楚地表明,回彈角的神經(jīng)網(wǎng)絡(luò)控制系統(tǒng),大大低于和接近原來的0.4至0.5度范圍這所有三種情況。即使如標稱t和m值,1.0毫米和0.10,分別使用的材料5,神經(jīng)網(wǎng)絡(luò)控制系統(tǒng)超越了閉環(huán)控制系統(tǒng)。
雖然神經(jīng)網(wǎng)絡(luò)系統(tǒng)需要生產(chǎn)足夠數(shù)量的實例實訓(xùn)網(wǎng)絡(luò)作為后續(xù)工作,但是帶來的好處是回彈量和最大應(yīng)變的控制。此外,還有額外的好處,神經(jīng)網(wǎng)絡(luò)控制系統(tǒng)在這個閉環(huán)控制策略,的這個的價值基與進行調(diào)整的材料4,表3相比,材料3與5的事件相比,為了使局部軌跡沖壓應(yīng)遵循準確。這就是說,作為材料和工藝參數(shù)的變化,取得與后續(xù)力軌跡密切合作是必要的但是也可能改變的。圖7還顯示本來材料4的實際沖壓力軌跡,如果k值是材料3和5,2.0中被使用。因此,當面臨著過程的很多變化和物質(zhì)需求,這個預(yù)先選定的閉環(huán)控制系統(tǒng)的實際收獲并不是很大。目前,行業(yè)中發(fā)現(xiàn)的應(yīng)力有隨時加強生產(chǎn)粘結(jié)劑力軌跡的能力,這是用于神經(jīng)網(wǎng)絡(luò)控制系統(tǒng),同時隨著一個不斷變化軌跡沖壓力,我們將需要更加有力的控制系統(tǒng)。
結(jié)論
本文提出了一種神經(jīng)網(wǎng)絡(luò)系統(tǒng),同時加強局部粘結(jié)劑軌跡,在模擬通道鋁成形中提出了控制回彈和最大應(yīng)變。神經(jīng)網(wǎng)絡(luò)被選用,源于其有處理高度線性問題的能力,并能發(fā)現(xiàn)金屬成形過程中不同的材料和工藝參數(shù)發(fā)生的變化。沖壓力的軌跡被確定為工藝參數(shù),提供了各種材料和工藝的變數(shù)最大的可供參考的偏差值。因此,多項式系數(shù)從曲線擬合的沖壓力軌跡被用來作為輸入的神經(jīng)網(wǎng)絡(luò)。結(jié)果表明,為加強局部粘結(jié)劑軌跡神經(jīng)網(wǎng)絡(luò)成功的提供了高粘結(jié)力(HBF)和沖壓位置(PD%)值,產(chǎn)生可接受的值回彈(ü),0.2至0.6度,最大應(yīng)變(e),8%至10%,得到最終產(chǎn)品時面臨著不同的材料應(yīng)力(k)+20%,+16%的應(yīng)變硬化指數(shù)(n),+25%的金屬片厚度(t),和+65%的摩擦系數(shù)(m)。又當Sunseri et al.(1996)提出的與閉環(huán)控制戰(zhàn)略在同一過程中,神經(jīng)網(wǎng)絡(luò)系統(tǒng)被證明在變化材料特性優(yōu)于閉環(huán)系統(tǒng)而進一步減少回彈,,如果有適當?shù)膶嵱?xùn)數(shù)據(jù),將被作為一個更加要有效的制度來執(zhí)行。
雖然這項工作進行了使用模擬,制定的方法可以很容易地延伸到實際的形成過程或?qū)嶒?,但是在未來將進行更多的實驗以驗證我們的需求。唯一的硬件要求是將一個有運算能力的CPU來測量沖壓力軌跡,并有能力改變一旦循環(huán)過程中生效的粘結(jié)劑。神經(jīng)網(wǎng)絡(luò)控制系統(tǒng)在一個金屬板料成形的進程中的材料工藝參數(shù)將有力的變化;因此,建立一致局部回彈量,對接下來的流程和客戶滿意度是至關(guān)重要的。
致謝
這項研究提供了部分經(jīng)費由美國國家科學(xué)基金會贈款#CMS-9622271和#DMI-9703249。
參考文獻:
ABAQUS,1997,User’s Manual,Theory Manual,Version 5.6,Hibbit, Karlsson,and Sorensen,Providence,RI.
Ayres,R.A.,1984,“SHAPESET:A Process to Reduce Sidewall Curl Springback in High-Strength Steel Rails,” Journal of Applied Metalworking, Vol.3,No.2,pp.127–34.
Cho,S.,Cho,Y.,and Yoon,S.,1997,“Reliable Roll Force Prediction in Cold Mill Using Multiple Neural Networks,”IEEE Transactions on Neural Networks,Vol.8,No.4,pp.874–82.
Di,S.,and Thomson,P.F.,1997,“Neural Network Approach for Prediction of Wrinkling Limit in Square Metal Sheet Under Diagonal Tension,”Journal of Testing and Evaluation, Vol.25,pp.74–81.
Elkins,K.L.,1994,“On-Line Angle Control for Small Radius Air Bending,”Carnegie Mellon University,UMI Dissertation Services,Ph.D.thesis.
Forcellese,A.,Gabrielli F.,and Ruffini,R.,1997,“Springback Controlin a Air Bending Process by Neural Network,”Submitted to Proc. III Convegno AITEM.
Graf,A.,and Hosford,W.,1993,“Plane-strain Tension Test of AL 2008-T4Sheets,” Proceedings SAE Congress Symposium on Sheet Forming, Detroit,MI.
Hishida,Y.,andWagoner,R., 1993,“Experimental Analysis of Blank Holding Force Control in Sheet Forming,”Journal of Materials and Manufacturing,Vol.2,pp.409–15.
Hornik,K.,Stinchcombe,M.,and White,H.,1989,“Multilayer Feedforward Networks are Universal Approximators,”Neural Networks,Vol. 2,pp.359–66.
Kinsey,B.,1998,“Process Control in Sheet Metal Forming,”Master’sthesis,Department of Mechanical Engineering,Northwestern University.
Kinsey,B.,and Cao,J.,1997,“An Experimental Study to Determine the Feasibility of Implementing Process Control to Reduce Part Variation in a Stamping Plant,” Sheet Metal Stamping: Development Applications, SAE Paper 970713, SP-1221, pp. 107–12.
Ruffini, R., and Cao, J., 1998, “Using Neural Network for Springback Minimization in a Channel Forming Process,”Developments in Sheet Metal Stamping,SAE Paper 98M-154,SP-1322,pp.77–85.
Rumelhart,D.E.and McClelland,J. L.,1986,“Learning InternalRepresentation by Error Propagation,”Parallel Distributed Processing,MIT Press,Cambridge,MA,Vol.1,Ch. 8.
Sunseri,M.,Cao,J.,Karafillis,A.P.,and oyce,M.C.,1996,“Accommodation of Springback Error in Channel Forming Using Active Binder Force Control:Numerical
Sunseri,M.,Cao,J.,Karafillis,A.P.,and oyce,M.C.,1996,“Accommodation of Springback Error in Channel Forming Using Active Binder Force Control: Numerical Simulations and Results,” ASME JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY, Vol.118,No.3,pp.426–35.
Widrow,B.and Lher,M.A.,1990,“30 Years of Adaptive Neural Networks: Perceptron,Madaline and Backpropagation,”Proceedings of the IEEE,Vol.78,No.9,pp.1415–41.
Yang M.,Shima, S.,and Watanabe,T.,1992,“Development of Control System Using Neural Network Combined with Deformation Model for an Intelligent V-Bending Process of Sheet Metals,” Proceedings of Japan/USA Symposium on Flexible Automation, ASME,Vol.2,pp.1485–90.
件2:外文原文(復(fù)印件)
Consistent and Minimal Springback Using a Stepped Binder Force Trajectory and Neural Network Control
Jian Cao Brad Kinsey
Department of Mechanical Engineering,Northwestern University, Evanston, IL 60208
Sara A. Solla
Department of Physics and Astronomy,Northwestern University, Evanston, IL 60208;
Department of Physiology,Northwestern University Medical School,Chicago, IL 60611
One of the greatest challenges of manufacturing sheet metal parts is to obtain consistent part dimensions. Springback, the elastic material recovery when the tooling is removed,is the major cause of variations and inconsistencies in the final part geometry. Obtaining a consistent and desirable amount of springback is extremely difficult due to the nonlinear effects and interactions between process and material parameters. In this paper, the exceptional ability of a neural network along with a stepped binder force trajectory to control springback angle and maximum principal strain in a simulated channel forming process is demonstrated. When faced with even large variations in material properties,sheet thickness, and friction condition, our control system produces a robust final part shape.
Introduction
Obtaining consistent and accurate part dimensions is crucial in today’s competitive manufacturing industry.Inconsistencies in part dimensions slow new product launches, increase changeover times, create difficulties in downstream processes,require extra quality assurance efforts,and decrease customer satisfaction and loyalty for the final product.In the sheet metal forming process,a major factor preventing accurate final part dimensions is springback in the material.Springback is the geometric difference between the part in its fully loaded condition,i.e., conforming to the tooling geometry,and when the part is in its unloaded, free state.For a complicated 3-D part,undesirable twist is another form of springback.The uneven distribution of stress through the sheet thickness direction and across the stamping in the loaded condition relaxes during unloading,thus producing springback.Factors that affect the amount of springback include variations in both process and material parameters,such as friction condition, tooling geometry,material properties,sheet thickness,and die temperature.Because controlling all of these variables in the manufacturing process is nearly impossible, springback,in turn,cannot be readily controlled.Adding to the difficulty is the fact that springback is a highly nonlinear effect; therefore, simulations and correcting methods are complicated. There has been a tremendous amount of research interest related to springback in recent years as is evident in proceedings of Society of Automotive Engineers,NUMIFORM,and NUMISHEET conferences.
Springback can be reduced through modifications to the forming process.Several researchers have proposed to use a stepped binder force trajectory to accomplish this objective (Ayres,1984;Hishida and Wagoner,1993;Sunseri et al.,1996).A stepped binder force trajectory is an instantaneous jump from a low binder force (LBF) value to a high binder force (HBF)at a specified percentage of the total punch displacement (PD%)(see Fig. 1).Sunseri et al.(1996)investigated springback in the Aluminum channel forming process shown in Fig. 2. Their work was conducted through experiments and simulations at specific val ues for process and material param-parameters.In a production environment, however, the amount of springback will deviate from the desired level due to variations in the process. Therefore, a control system that accommodates variations in process parameters is required.
In recent years, many research groups have investigated the use of artificial neural networks to control sheet metal forming processes. Metal forming is an ideal candidate for neural network control due to the nonlinear effects and interactions of the process parameters. Cho et al. (1997) used a neural network to predict the force in cold rolling, and Di and Thomson (1997) predicted the wrinkling limit in square metal sheets under diagonal tension. Among others (Elkins, 1994 and Yang et al., 1992), Forcellese et al. (1997) used a neural network to control springback in a 60 deg aluminum V-punch air bending process. Their system was trainedusing experimentally obtained examples consisting of five parameters from the punch force trajectory, an off-line measurement of sheet thickness, and the target bend angle as the inputs into the neural network and the punch displacement as the output. In another research project, Ruffini and Cao (1998) proposed to use a neural network to control springback angle in a channel forming process with punch force trajectory as the sole source for identifying the process variations and adjusting the HBF used in Sunseri et al. (1996). Preliminary results showed this approach to be promising
In this paper, the springback of an aluminum channel is controlled via a stepped binder force trajectory and neural network control. The neural network determines the HBF and PD% of the stepped binder force trajectory. Punch force trajectory was identified as the key parameter that reflects variations in material properties, sheet thickness, and friction coefficient. Therefore, four polynomial coefficients from curve fitting the punch force trajectory were used as inputs into the neural network. Figure 3 shows a flowchart of our proposed control system for this application. Despite large variations in material properties, sheet thickness (t), and friction coefficient (m), the springback angle (u) was maintained between 0.2 and 0.6 deg and the maximum strain (e) was limited to between 8% and 10%. Finally, a comparison with the closed-loop control method proposed by Sunseri et al. (1996) is included to show the benefits of our control method. While only numerical simulation results are presented here, the control system will be physically implementation in the future to verify improvement claims
.
A simple geometry to investigate spr ingback is a channel forming process. Therefore, the same aluminum channel forming process used by Sunseri et al. (1996) is employed here (see Fig. 2). First, the effect of constant binder force (CBF) on springback was evaluated. As the CBF was increased while all other process and material parameters were held constant in our simulations, the springback angle u, was reduced as shown graphically in Fig. 4 and physically in Fig. 5. However, the increased binder force caused a subsequent increase in the maximum strain1 in the material, solid line in Fig. 4, to levels that exceed the maximum stretchability of aluminum (Graf and Hosford, 1993). By utilizing a stepped binder force trajectory, moderate maximum strain levels in the sidewall were obtained while reducing springback in the process as demonstrated in the table of Fig. 1.
To produce a robust process when faced with deviations in the friction coefficient, Sunseri et al. (1996) implemented closed-loop variable binder force control to follow the punch force trajectory obtained from the stepped binder force case with nominal process conditions. This control method was able to produce consistent springback levels when the friction coefficient was varied from a value of 0.1, for the nominal case, to 0.25. However, whether this methodology could withstand variations of other parameters such as material properties and sheet thickness was not determined.
Finite Element Model.
A commercial Finite Element Analysis package (ABAQUS, 1997) was used for our numerical simulations of the channel forming process. Since the problem is close to a plane-strain condition and is symmetric, only one sixteenth of the width and half of the length of the entire blank (220 mm 3 46 mm) was modelled. The binder, the die, and the punch were modelled as three separate rigid surfaces. Each surface was modelled using four-node interface elements (ABAQUS type IRS4), and a Coulomb friction law was assumed. Our blank mesh had an uneven distribution of 40 four-node, reduced integration shell elements (ABAQUS type S4R) with a more dense concentration of elements where the blank contacted the punch and binder corner radii. Boundary conditions were specified to create a plane-strain condition. The material was modelled to be isotropic, elasto-plastic following the von Mises yield criterion and isotropic strain hardening. The elastic properties were the Young’s modulus, E, of 70 GPa and Poisson’s ratio, n, of 0.3. The plastic behavior of the sheet material was modelled using a power law relation (s 5 Ke n). Our nominal material, denoted Material 1, had a material strength coefficient, K, of 528 MPa and a strain hardening exponent, n, of 0.265.
Proposed Control System
In a channel forming process, spri