亚洲免费av电影一区二区三区,日韩爱爱视频,51精品视频一区二区三区,91视频爱爱,日韩欧美在线播放视频,中文字幕少妇AV,亚洲电影中文字幕,久久久久亚洲av成人网址,久久综合视频网站,国产在线不卡免费播放

        ?

        基于多層感知人工神經(jīng)網(wǎng)絡(luò)的執(zhí)行機(jī)構(gòu)末端綜合定位

        2016-04-09 03:16:41胡燕祝李雷遠(yuǎn)北京郵電大學(xué)自動(dòng)化學(xué)院北京100876
        關(guān)鍵詞:機(jī)械臂控制模型

        胡燕祝,李雷遠(yuǎn)(北京郵電大學(xué)自動(dòng)化學(xué)院,北京100876)

        ?

        基于多層感知人工神經(jīng)網(wǎng)絡(luò)的執(zhí)行機(jī)構(gòu)末端綜合定位

        胡燕祝,李雷遠(yuǎn)
        (北京郵電大學(xué)自動(dòng)化學(xué)院,北京100876)

        摘要:非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)的雅可比矩陣和連桿坐標(biāo)系往往難以確定,導(dǎo)致任務(wù)空間的定位性難以分析。論文提出并證明綜合型定位方法的充分必要條件,即完成多種特別定位任務(wù)的充要條件;用反向傳播的多層感知人工神經(jīng)網(wǎng)絡(luò)(MLP, multilayer perceptron neural network)求解逆運(yùn)動(dòng)學(xué)模型,在笛卡爾空間,把執(zhí)行機(jī)構(gòu)D-H(denavit-hartenberg)參數(shù)作為訓(xùn)練集,對(duì)神經(jīng)網(wǎng)絡(luò)進(jìn)行訓(xùn)練;定義一個(gè)函數(shù),判斷執(zhí)行機(jī)構(gòu)定位到目標(biāo)點(diǎn)的性能,即可定位性。經(jīng)仿真驗(yàn)證,神經(jīng)網(wǎng)絡(luò)求解逆運(yùn)動(dòng)學(xué)模型,較傳統(tǒng)方法縮短了計(jì)算時(shí)間,計(jì)算效率提高20%,精度提高2.4%,可定位性最小值為0.96,最優(yōu)運(yùn)動(dòng)學(xué)函數(shù)值4.0349×1014。

        關(guān)鍵詞:機(jī)械化;控制;模型;機(jī)械臂;神經(jīng)網(wǎng)絡(luò);逆運(yùn)動(dòng)學(xué)

        胡燕祝,李雷遠(yuǎn).基于多層感知人工神經(jīng)網(wǎng)絡(luò)的執(zhí)行機(jī)構(gòu)末端綜合定位[J].農(nóng)業(yè)工程學(xué)報(bào),2016,32(01):22-29.doi:10.11975/j.issn.1002-6819.2016.01.003 http://www.tcsae.org

        Hu Yanzhu, Li Leiyuan.Series actuator end integrated positioning analysis based-on multilayer perceptron neural network [J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE), 2016, 32(01): 22-29.(in Chinese with English abstract)doi:10.11975/j.issn.1002-6819.2016.01.003 http://www.tcsae.org

        0 引言

        標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)較為通用,但也不能保證執(zhí)行最優(yōu)的任務(wù),工業(yè)執(zhí)行機(jī)構(gòu)只能重復(fù)執(zhí)行一系列給定任務(wù)。由此,特殊任務(wù)或帶有優(yōu)化任務(wù)的非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)是急需的,如個(gè)性化精密制造、裝配和包裝等。若要完成這些個(gè)性化任務(wù),則必須對(duì)執(zhí)行機(jī)構(gòu)的定位性或可到達(dá)。

        空間性進(jìn)行分析。本文分為3步,第1步,確定D-H (denavit-hartenberg)參數(shù),建立正運(yùn)動(dòng)學(xué)模型[1],證明綜合定位任務(wù)的充要條件;第2步,人工神經(jīng)網(wǎng)絡(luò)(ANNs,artificial neural network)求解逆運(yùn)動(dòng)學(xué);第3步,建立定位性函數(shù),評(píng)價(jià)定位任務(wù)性能。

        非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)定位性分析方法可分為3種,幾何分析法、參數(shù)優(yōu)化分析法和基于任務(wù)分析法。幾何分析法有其局限性,不能擴(kuò)展和應(yīng)用到棱形連桿結(jié)構(gòu),同時(shí)不滿足多任務(wù)定位需求;參數(shù)優(yōu)化法的主要缺點(diǎn)是受限于關(guān)節(jié)自由度和關(guān)節(jié)極限位置等;基于任務(wù)的分析方法應(yīng)用先驗(yàn)知識(shí),生成齊次變換矩陣和定位動(dòng)態(tài)參數(shù);Paredis和Kholsa[2]應(yīng)用基于任務(wù)分析方法產(chǎn)生D-H參數(shù),并對(duì)6-DOF(degree of freedom)標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)末端進(jìn)行定位分析;Kholsa等[3]在任務(wù)分析法基礎(chǔ)上,提出任務(wù)描述概念,進(jìn)行運(yùn)動(dòng)學(xué)建模、定位規(guī)劃和定位控制等。本文在任務(wù)分析法的基礎(chǔ)上,對(duì)非結(jié)構(gòu)化執(zhí)行機(jī)構(gòu)進(jìn)行定位問題描述、建立綜合充要條件和定位性評(píng)價(jià)。

        雅可比矩陣在執(zhí)行機(jī)構(gòu)的運(yùn)動(dòng)學(xué)分析中具有重要地位,機(jī)器人的分離速度控制、靜力分析和靈活性和可操作度分析等都要用到機(jī)器人的雅可比矩陣[4]。雅可比矩陣的構(gòu)造方法有矢量積法、微分變換法、力和力矩遞推法和速度遞推法?;谘趴杀染仃嚹軌蚍治龇€(wěn)定性(特征根)、雅可比矩陣的奇異性用矩陣的秩來描述,滿秩時(shí),可進(jìn)行奇異值分解;不滿秩時(shí),執(zhí)行機(jī)構(gòu)處于奇異位形。執(zhí)行機(jī)構(gòu)的靈活性與運(yùn)動(dòng)學(xué)逆解的精度與雅可比矩陣奇異值有關(guān)。同時(shí),雅可比矩陣是由關(guān)節(jié)速度映射到執(zhí)行機(jī)構(gòu)末端速度,因此也是構(gòu)成綜合定位[5]充要條件的關(guān)鍵因素之一。

        已知執(zhí)行機(jī)構(gòu)末端在操作空間中的位姿,利用逆運(yùn)動(dòng)學(xué)解出關(guān)節(jié)角度。傳統(tǒng)的逆運(yùn)動(dòng)學(xué)求解方法有幾何法、迭代法、代數(shù)法和Particle Swarm優(yōu)化法等。其封閉解也不是萬能的,只有在執(zhí)行機(jī)構(gòu)的雅可比矩陣為滿秩時(shí)才成立。其次,逆運(yùn)動(dòng)學(xué)方程通常沒有獨(dú)一無二的解,因?yàn)樵陉P(guān)節(jié)空間中,執(zhí)行機(jī)構(gòu)存在多個(gè)位姿,使之定位到任務(wù)空間中的目標(biāo)位置。逆運(yùn)動(dòng)學(xué)求解時(shí),應(yīng)避免奇異性域,奇異性即是執(zhí)行機(jī)構(gòu)的2個(gè)或更多個(gè)旋轉(zhuǎn)軸共線引起的不可預(yù)測(cè)的運(yùn)動(dòng)和速度。奇異性的存在影響執(zhí)行機(jī)構(gòu)末端的定位,人工神經(jīng)網(wǎng)絡(luò)逆運(yùn)動(dòng)學(xué)求解方法可以盡量消除奇異域[6]。在網(wǎng)絡(luò)訓(xùn)練和學(xué)習(xí)時(shí),通過Levenberg-Marquart (LM)算法優(yōu)化均方差(MSE)。以奇異點(diǎn)處的笛卡爾坐標(biāo)作為測(cè)試集,以執(zhí)行機(jī)構(gòu)笛卡爾坐標(biāo)和D-H參數(shù)作為訓(xùn)練集。

        為了判斷非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)末端能否定位到給定點(diǎn),及到達(dá)給定點(diǎn)的收斂程度,引入可到達(dá)性函數(shù),即可定位性函數(shù)[7-8]??啥ㄎ恍院瘮?shù)反映非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)運(yùn)動(dòng)學(xué)性能。

        1 運(yùn)動(dòng)學(xué)分析

        首先,對(duì)論文中用到的一些空間參數(shù)和變量進(jìn)行說明,ΓT表示任務(wù)空間,ΓQ關(guān)節(jié)空間,r1表示大臂長(zhǎng)度,r2表示小臂長(zhǎng)度,r3腕部長(zhǎng)度,ΓQC受限的關(guān)節(jié)空間,qi各關(guān)節(jié)運(yùn)動(dòng)矢量,ξE執(zhí)行機(jī)構(gòu)末端位姿,ΓC配置空間,0AE基座相對(duì)于執(zhí)行機(jī)構(gòu)末端的變換矩陣,ΓW可到達(dá)或可定位空間,ΓCW受關(guān)節(jié)限制的可到達(dá)或可定位空間。

        非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)具有5個(gè)自由度3連桿,如圖1a。5個(gè)旋轉(zhuǎn)軸分別為腰部旋轉(zhuǎn)軸,大臂俯仰軸,小臂俯仰軸,腕部俯仰軸和腕部旋轉(zhuǎn)軸。5關(guān)節(jié)分別為腰部旋轉(zhuǎn)關(guān)節(jié),大臂俯仰關(guān)節(jié),小臂俯仰關(guān)節(jié),腕部俯仰關(guān)節(jié)和腕部旋轉(zhuǎn)關(guān)節(jié)。角度儀測(cè)量各關(guān)節(jié)極限角度,以車體水平面作為參考平面,如表1所示。

        圖1 非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)及坐標(biāo)系Fig.1 Non standardized execution mechanism and DH parameter coordinate system

        表1 執(zhí)行機(jī)構(gòu)工作空間和最大轉(zhuǎn)速Table 1 Working space and maximum speed of actuator

        {a,α,d}表示旋轉(zhuǎn)型連桿參數(shù),{a,α,θ}表示棱型連桿的參數(shù)。標(biāo)準(zhǔn)D-H參數(shù)存在限制性,其關(guān)節(jié)必須繞z軸旋轉(zhuǎn),連桿在x軸上產(chǎn)生位移。在配置空間ΓC中,5(N)自由度執(zhí)行機(jī)構(gòu)具有15(3N)個(gè)配置參數(shù),因此D-H參數(shù)集構(gòu)成了一個(gè)15維的ΓC空間,表達(dá)如下:

        正運(yùn)動(dòng)學(xué)模型可用下列函數(shù)表示:

        表示執(zhí)行機(jī)構(gòu)末端的運(yùn)動(dòng)矢量,q為各關(guān)節(jié)運(yùn)動(dòng)矢量。正運(yùn)動(dòng)學(xué)變換矩陣是1個(gè)4×4方陣。n,b和t分別表示執(zhí)行機(jī)構(gòu)末端在x,y和z軸的方向,p表示執(zhí)行機(jī)構(gòu)末端的位置(笛卡爾坐標(biāo)系)。

        在空間ΓCW內(nèi),執(zhí)行機(jī)構(gòu)末端的空間坐標(biāo)為:

        2 綜合定位充要條件

        非標(biāo)準(zhǔn)化5軸執(zhí)行機(jī)構(gòu)完成綜合定位任務(wù)的充要條件為:能夠找到所有的D-H參數(shù)滿足(DH,q)=p且rank(Jacobian(q))=5。證明如下。

        充分性:若非標(biāo)準(zhǔn)化5軸執(zhí)行機(jī)構(gòu)可以完成綜合定位任務(wù),則能夠找到所有的D-H參數(shù)滿足?p∈ΓT,?q∈ΓQCf(DH,q)=p且rank(Jacobian(q))=5。

        眾所周知,DH∈ΓC,ΓCW?ΓW,ΓQC?ΓQ,設(shè)P是ΓT內(nèi)的一個(gè)點(diǎn)集,

        ΓT內(nèi)的每一個(gè)點(diǎn)含有6個(gè)維度,分別用執(zhí)行機(jī)構(gòu)末端的位置和方向進(jìn)行定義,如下:

        對(duì)于5自由度的非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu),其ΓQ維數(shù)為5,所以,關(guān)節(jié)向量為:

        每個(gè)關(guān)節(jié)在ΓQC內(nèi),都有其極限位姿,上限位姿和下限位姿約束為:

        定義ΓW為無關(guān)節(jié)限制時(shí)非標(biāo)準(zhǔn)化執(zhí)行機(jī)構(gòu)末端能夠達(dá)到或定位到的世界坐標(biāo)系內(nèi)所有點(diǎn)的集合[11-15]。根據(jù)正運(yùn)動(dòng)學(xué)原理(2)得到映射,

        同理,在ΓWS內(nèi),且ΓCW?ΓW,ΓQC?ΓQ,

        再由(1)式,得

        由于執(zhí)行機(jī)構(gòu)任務(wù)空間存在奇異性,通過雅可比矩陣予以避免。對(duì)(2)求導(dǎo),

        執(zhí)行機(jī)構(gòu)雅可比矩陣[16-17]是1個(gè)6×N矩陣,N為關(guān)節(jié)數(shù)。為了避免奇異性,rank(J(DH,q))=5。推導(dǎo)出結(jié)論:存在D-H參數(shù)滿足且rank(Jacobian(q))=5。

        必要性:如果能夠找到D-H參數(shù)滿足?p∈ΓT,?q∈且rank(Jacobian(q))=5,則非標(biāo)準(zhǔn)化5軸執(zhí)行機(jī)構(gòu)能夠完成綜合定位任務(wù)。

        3 人工神經(jīng)網(wǎng)絡(luò)逆運(yùn)動(dòng)學(xué)求解

        給定任務(wù)空間執(zhí)行機(jī)構(gòu)末端的位置坐標(biāo)(X,Y,Z),通過ANNs確定執(zhí)行機(jī)構(gòu)各關(guān)節(jié)角度θ1,θ2,θ3,θ4和θ5。神經(jīng)網(wǎng)絡(luò)采用前饋多層感知(MLP,multilayer perceptron neural network),把執(zhí)行機(jī)構(gòu)的笛卡爾坐標(biāo)和D-H參數(shù)作為訓(xùn)練集[18-20],不斷更新神經(jīng)網(wǎng)絡(luò)權(quán)值,使均方誤差參數(shù)(MSE, mean square error)達(dá)到最?。或?yàn)證集,確定網(wǎng)絡(luò)結(jié)構(gòu)或者控制模型復(fù)雜程度的參數(shù),當(dāng)泛化值停止改善時(shí),停止訓(xùn)練;訓(xùn)練后,用測(cè)試集對(duì)神經(jīng)網(wǎng)絡(luò)性能進(jìn)行獨(dú)立測(cè)試。

        MLP前饋反向傳播網(wǎng)絡(luò)用權(quán)值(w)和偏移量(b)表示如下:

        xi為網(wǎng)絡(luò)輸入,即(X,Y,Z);wi為每個(gè)輸入的權(quán)值;b為偏移量;S為輸出,即(θ1,θ2,θ3,θ4,θ5);神經(jīng)網(wǎng)絡(luò)架構(gòu),如圖2所示。

        圖2 神經(jīng)網(wǎng)絡(luò)架構(gòu)Fig.2 Neural network architecture

        采用Levenberg-Marquardt(LM)算法[21-23]訓(xùn)練MLP網(wǎng)絡(luò),通過調(diào)整學(xué)習(xí)速率,獲得最優(yōu)權(quán)值wi,并且使均方誤差(MSE)取得最小值。LM算法以Newton算法和梯度下降算法為基礎(chǔ),取誤差函數(shù)Ei的1階導(dǎo)數(shù),使全局誤差最小[24]。表達(dá)式為,

        d為Ei的1階導(dǎo)數(shù),ds為Ei的2階導(dǎo)數(shù),e為自然對(duì)數(shù)函數(shù),λ為阻尼因子。均方誤差函數(shù)的表達(dá)式如下,

        Ei為第i個(gè)輸入數(shù)據(jù)的誤差,n為輸入數(shù)據(jù)數(shù)量。

        MLP神經(jīng)網(wǎng)絡(luò)采用有監(jiān)督學(xué)習(xí),包含3個(gè)輸入,帶有20個(gè)神經(jīng)元的隱層,5個(gè)輸出。隱層具有tansigmoid激勵(lì)函數(shù),范圍[-1,1];輸出層具有pureline線性激勵(lì)函數(shù)。

        神經(jīng)網(wǎng)絡(luò)訓(xùn)練集、驗(yàn)證集和測(cè)試集是歸一化后的數(shù)據(jù),要求取值范圍在(0,1)內(nèi),所以采用歸一化函數(shù)f(x)=1/(1+e-x)。證明如下,令x∈(-∞,∞),

        4 可定位性函數(shù)

        神經(jīng)網(wǎng)絡(luò)訓(xùn)練出來的模型,進(jìn)行逆運(yùn)動(dòng)學(xué)求解時(shí)[25],不存在復(fù)數(shù)解,因此,不考慮復(fù)數(shù)解情況。建立一個(gè)可定位性函數(shù),判定執(zhí)行機(jī)構(gòu)末端能否按照需要的方向到達(dá)任務(wù)空間中目標(biāo)點(diǎn)[26-29]。利用歸一化模型,評(píng)價(jià)ΓT空間中任意點(diǎn)的可定位性,

        g為點(diǎn)p對(duì)應(yīng)所有運(yùn)動(dòng)學(xué)逆解的數(shù)量,函數(shù)reachability(DH)取值范圍為[0,1]。當(dāng)?。?,1)時(shí),逆運(yùn)動(dòng)學(xué)解中至少有一組受關(guān)節(jié)限制而無法定位到;當(dāng)取0時(shí),最優(yōu)解中至少有一個(gè)關(guān)節(jié)角度超過極限位置;當(dāng)取1時(shí),最優(yōu)解中所有關(guān)節(jié)都處于中間位置,無論哪一組解都可以定位到目標(biāo)點(diǎn)[30-31]。最優(yōu)逆運(yùn)動(dòng)學(xué)解選取,即執(zhí)行機(jī)構(gòu)的最優(yōu)配置,函數(shù)如下,

        其中,J(DH,q1)是6×5的矩陣,JT(DH,q1))J(DH,q1)為5×5矩陣,評(píng)價(jià)ΓQ內(nèi)NUM個(gè)點(diǎn)的可定位性能,

        評(píng)價(jià)ΓT內(nèi)定位到NUM個(gè)點(diǎn)的運(yùn)動(dòng)學(xué)性能,有

        取目標(biāo)函數(shù)的最大值,是為了找到目標(biāo)點(diǎn)處的最合適的速度變換矩陣,即雅可比矩陣。雅可比矩陣是由關(guān)節(jié)速度到執(zhí)行機(jī)構(gòu)末端速度的映射,能夠反映執(zhí)行機(jī)構(gòu)的運(yùn)動(dòng)學(xué)性能。

        5 試驗(yàn)與分析

        此執(zhí)行機(jī)構(gòu)的標(biāo)準(zhǔn)D-H參數(shù)如表2。DH參數(shù)對(duì)應(yīng)的正運(yùn)動(dòng)學(xué)變換矩陣為,

        所以,初始狀態(tài)末端空間坐標(biāo)為(433.5,0,-255.4)。由正運(yùn)動(dòng)學(xué)推導(dǎo)末端運(yùn)動(dòng)區(qū)域,如圖3所示,圖中各個(gè)點(diǎn)表示執(zhí)行機(jī)構(gòu)末端所能達(dá)到的空間坐標(biāo),近似球形。執(zhí)行機(jī)構(gòu)末端運(yùn)動(dòng)范圍構(gòu)成ΓT,所以?p∈ΓT,?q∈ΓQCf(DH,q)=p。此執(zhí)行機(jī)構(gòu)的雅可比矩陣是一個(gè)6*5矩陣,初始狀態(tài)下DH參數(shù)對(duì)應(yīng)的雅可比矩陣為,

        表2 標(biāo)準(zhǔn)D-H參數(shù)Table 2 Standard D-H parameters

        圖3 末端運(yùn)動(dòng)范圍Fig.3 Manipulator end motion range

        利用傳統(tǒng)方法計(jì)算逆運(yùn)動(dòng)學(xué)解,共2 000組數(shù)據(jù),在ΓT內(nèi),(X,Y,Z)數(shù)據(jù)隨機(jī)生成,如圖3中的點(diǎn)集,逆運(yùn)動(dòng)學(xué)解為(Θ1,Θ2,Θ3,Θ4,Θ5),并將(X,Y,Z,Θ1,Θ2,Θ3,Θ4,Θ5)數(shù)據(jù)按列歸一化,歸一化函數(shù)為f(x)=1/(1+e-x)。1 400組(X,Y,Z,Θ1,Θ2,Θ3,Θ4,Θ5)作為訓(xùn)練樣本,構(gòu)成1400*8維矩陣,用試湊法確定最佳隱層節(jié)點(diǎn)數(shù),先設(shè)置較少的隱節(jié)點(diǎn)訓(xùn)練網(wǎng)絡(luò),然后逐漸增加隱節(jié)點(diǎn)數(shù),進(jìn)行訓(xùn)練,確定網(wǎng)絡(luò)誤差最小時(shí)對(duì)應(yīng)的隱節(jié)點(diǎn)數(shù)為20;最大訓(xùn)練次數(shù)為28 302,訓(xùn)練目標(biāo)誤差設(shè)置為1×10-4,動(dòng)量因子常數(shù)為1×10-4,學(xué)習(xí)速率為1×10-5,學(xué)習(xí)速率增加比率為5,學(xué)習(xí)速率減少比率0.5;引起訓(xùn)練結(jié)束的條件是Validation Checks為20,訓(xùn)練時(shí)間47 s,迭代30次,均方誤差為5.96×10-5,梯度值為0.028 8;500組作為驗(yàn)證樣本,維數(shù)500×8,100組測(cè)試樣本,維數(shù)100×8。訓(xùn)練樣本和測(cè)試樣本相互獨(dú)立。θ1到θ5的網(wǎng)絡(luò)輸出和理論計(jì)算值之間的誤差Err=網(wǎng)絡(luò)估計(jì)值-理論計(jì)算值,如圖4,θ1的估計(jì)誤差最大約為3.7,θ2的最大估計(jì)誤差約為3.1,θ3的最大估計(jì)誤差約為3.5,θ4的最大估計(jì)誤差約為3.3,θ5的最大估計(jì)誤差約為4.5,所有實(shí)際誤差均在5以內(nèi),能夠滿足實(shí)際需求。

        采用神經(jīng)網(wǎng)絡(luò)解逆運(yùn)動(dòng)學(xué),需要在計(jì)算時(shí)間和精度之間進(jìn)行權(quán)衡,往往是縮短了計(jì)算時(shí)間,而精度卻不理想。傳統(tǒng)算法計(jì)算時(shí)間為1.2 s,神經(jīng)網(wǎng)絡(luò)訓(xùn)練好后,計(jì)算時(shí)間為0.9 s,效率提高了20%。Luv Aggarwal等用神經(jīng)網(wǎng)絡(luò)求解精度為87.5%,論文神經(jīng)網(wǎng)絡(luò)求解精度為89.9%,提高了2.4%。因此,在實(shí)時(shí)性要求不高而精度要求較高時(shí),應(yīng)該采用傳統(tǒng)計(jì)算方法,而在實(shí)時(shí)性要求較高而精度要求不高且系統(tǒng)為非線性時(shí),可以選擇神經(jīng)網(wǎng)絡(luò)求解方法。

        表3 定位到(41.4, 89.0, 104.5)時(shí)第1組解Table 3  First group solutions of positioning to(41.4, 89.0, 104.5)

        圖4 網(wǎng)絡(luò)估計(jì)值與理論計(jì)算值的誤差Fig.4 Error between estimated value and calculated value

        在ΓT內(nèi),?p∈ΓT,設(shè)定位點(diǎn)p=(41.4,89.0,104.5),逆運(yùn)動(dòng)學(xué)解如表3和表4,共2組解。由式(18),可計(jì)算每一組逆運(yùn)動(dòng)學(xué)解的可定位性值,如圖5(a),用紅色’.’表示,軸表示第幾組逆運(yùn)動(dòng)學(xué)解,y軸表示對(duì)應(yīng)的定位性值。第21組逆運(yùn)動(dòng)學(xué)解的可定性函數(shù)值最小,因此在p點(diǎn)處的可定位性為reachability(DH)=0.96,藍(lán)色’o’表示;最優(yōu)運(yùn)動(dòng)學(xué)性能為configuration(DH)=4.034 9×1014,1~21組解的最優(yōu)運(yùn)動(dòng)學(xué)性能如圖5(b)所示,用紅色’.’表示,x軸表示第幾組逆運(yùn)動(dòng)學(xué)解,y軸表示對(duì)應(yīng)的最優(yōu)運(yùn)動(dòng)學(xué)性能值,藍(lán)色’o’表示最優(yōu)運(yùn)動(dòng)學(xué)性能值中最大值。從圖中可以看出,第12組逆運(yùn)動(dòng)學(xué)解(21.61,125.73,108.42,99.41,0)可以達(dá)到最優(yōu)運(yùn)動(dòng)學(xué)。

        表4 定位到(41.4, 89.0, 104.5)時(shí)第2組解Table 4 Second group solutions of positioning to(41.4, 89.0, 104.5)

        圖5 點(diǎn)的可定位性及運(yùn)動(dòng)學(xué)性能分布Fig.5  Distribution of localization and kinematic performance for a point

        6 結(jié)論

        為滿足用戶個(gè)性化需求,需對(duì)執(zhí)行機(jī)構(gòu)的結(jié)構(gòu)重新設(shè)計(jì),提出可定位性充要條件,在滿足精度要求的前提下,盡量提高執(zhí)行機(jī)構(gòu)逆運(yùn)動(dòng)學(xué)求解效率,并對(duì)可定位性定量分析,其實(shí)質(zhì)是基于可定位性的最優(yōu)配置。論文對(duì)可定位性充要條件進(jìn)行證明,計(jì)算執(zhí)行機(jī)構(gòu)末端任務(wù)空間的運(yùn)動(dòng)范圍,訓(xùn)練基于MLP的神經(jīng)網(wǎng)絡(luò)并逆運(yùn)動(dòng)學(xué)求解,效率提高了20%,精度提高2.4%,并舉例計(jì)算p=(41.4,89.0,104.5)時(shí),可定位性值reachability(DH)=0.96,configuration(DH)=4.039×1014。接下來工作完成一整套環(huán)形或平面多點(diǎn)等任務(wù)的最低功耗研究。

        [參考文獻(xiàn)]

        [1]張曉燕,孫劍橋,丁千.5自由度機(jī)械臂的運(yùn)動(dòng)學(xué)分析與時(shí)滯控制[J].西安交通大學(xué)學(xué)報(bào),2013,10(47);103-108.Zhang Xiaoyan, Sun Jianqiao, Ding Qian.Kinematics analysis and time delayed control of a five-degree of freedom manipulator [J].Journal of Xi’An Jiaotong University, 2013, 10(47): 103-108.(in Chinese with English abstract)

        [2] Paredis C J, Khosla P K.Kinematic design of serial link manipulators from task specifications[J].IntJ Robotics Res, 1993, 12(3): 274-87.

        [3] Khosla P K.A multi-population genetic algorithm and its application to design of manipulators[J].In: Proc.IEEE/RSJ int conf on intell robots and syst, 1992, 3(24): 7-10.

        [4]李誠,謝志江,倪衛(wèi),等.六自由度裝校機(jī)器人雅可比矩陣的建立及奇異性分析[J].中國機(jī)械工程,2012,5(11):1165-1169.Li Cheng, Xie Zhijiang, Ni Wei, et al.Establishment of jacobian matrix and singularity analysis of a 6-DOF installing calibrating robot[J].China Mechanical Engineering, 2012, 5(11): 1165-1169.(in Chinese with English abstract)

        [5] Sarosh P, Tarek S.Task based synthesis of serial manipulators[J].Journal of Advanced Research, 2015, 39(8):2-7.

        [6] Luv A, Kush A, Ruth J U.Use of artificial neural networks for the development of an inverse kinematic solution and visual identification of singularity zone [J].Variety Management in Manufacturing, 2014,8(10):812-817.

        [7] Tian T X, Jiang H Z, Tong Z Z, et al.An internal parameter identification method of eliminating system damping effect for a six-degree-of-freedom parallel manipulator[J].Chinese Journal of Aeronautics, 2015, 12(24): 450-457.

        [8] Monika Z F, Ana M D.Complexity analysis for calculating the jacobian matrix of 6-DOF reconfigurable machines[J].Variety Management in Manufacturing, 2014, 2(19): 218-223.

        [9]任子武,王振華,李娟,等.基于混合類電磁機(jī)制算法的機(jī)械臂逆運(yùn)動(dòng)學(xué)解[J].機(jī)械工程學(xué)報(bào),2014,17(48):21-27.Ren Ziwu, Wang Zhenhua, Li Juan, et al.Inverse kinematics solution for robot manipulator based on hybrid electromagnetism-like mechanism algorithm [J].Journal ofMechanical Engineering, 2014, 17(48):21-27.(in Chinese with English abstract)

        [10]高自成,李立君,李昕,等.齒梳式油茶果采摘機(jī)采摘執(zhí)行機(jī)構(gòu)的研制與試驗(yàn)[J].農(nóng)業(yè)工程學(xué)報(bào),2013,10(27):38-44.Gao Zicheng, Li Lijun, Li Xin, et al.Development and test of picking actor in oil-tea camellia fruit picking machine of tooth comb type[J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE), 2013, 10(27): 38-44.(in Chinese with English abstract)

        [11]章德賓,徐家鵬,許建軍,等.基于監(jiān)測(cè)數(shù)據(jù)和BP神經(jīng)網(wǎng)絡(luò)的食品安全預(yù)警模型[J].農(nóng)業(yè)工程學(xué)報(bào),2010, 1(6):55-62.Zhang Debin, Xu Jiapeng, Xu Jianjun, et al.Model for food safety warning based on inspection data and BP neural network [J].Transactions of the Chinese Society of Agricultural (Transactions of the CSAE), 2010,1(6):55-62.(in Chinese with English abstract)

        [12] Yong J K, Shan B C, Sang B K, et al.A stiffness-adjustable hyper redundant manipulator using a variable neutral-line mechanism for minimally invasive surgery [J].IEEE Transactions on Robotics, 2014,2(30): 382-394.

        [13] Matteo C P, Luca C, Giacomo P, et al.Analysis and design of a reconfigurable 3-DoF parallel manipulator for multimodal tasks [J].IEEE/ASME Transactions on Mechatronics, 2015, 4(20): 1975-1984.

        [14] Janne K, Jouni M.Stability-guaranteed force-sensorless contact force/motion control of heavy-duty hydraulic manipulators [J].IEEE Transactions on Robotics, 2015, 4(31): 918-934.

        [15] Rong J W, Rajkumar M.Design of fuzzy-neural-networkinherited back stepping control for robot manipulator including actuator dynamics [J].IEEE Transactions on Fuzzy Systems, 2014, 4(22): 709-721.

        [16] Ahmet D, Koksal E.Trajectory tracking control for a 3-DOF parallel manipulator using fractional-order control [J].IEEE Transactions on Industrial Electronics, 2014, 7(61): 3417-3423.

        [17] Sehoon O, Kyoungchul K.Two-degree-of-freedom control of a two-link manipulator in the rotating coordinate system[J].IEEE Transactions on Industrial Electronics, 2015, 9(62): 5598-5607.

        [18] Kim S M.Lumped element modeling of a flexible manipulator system[J].IEEE/ASMETransactionson Mechatronics,2015,2(20): 967-973.

        [19] Chu Z Y, Cui J, Sun F C.Fuzzy adaptive disturbance-observerbased robust tracking control of electrically driven free-floating space manipulator[J].IEEE Systems Journal, 2014, 2(8):343-351.

        [20] Luca C, Massimo C, Giacomo P.Analysis of kinematics and reconfigurability of a spherical parallel manipulator[J].IEEE Transactions on Robotics, 2014, 6(30):1541-1546.

        [21]吳玉香,王聰.基于確定學(xué)習(xí)的機(jī)器人任務(wù)空間自適應(yīng)神經(jīng)網(wǎng)絡(luò)控制[J].自動(dòng)化學(xué)報(bào),2013,6(39):806-815.Wu Yuxiang, Wang Cong.Deterministic learning based adaptive network control of robot in task space[J].Acta Automatica Sinica, 2013, 6(39): 806-815.(in Chinese with English abstract)

        [22] Mai T L, Wang Y N.Adaptive force/motion control system based on recurrent fuzzy wavelet CMAC neural networks for condenser cleaning crawler - type mobile manipulator robot[J].IEEE Transactions on Control systems Technology, 2014, 5(22):1973-1981.

        [23] Long J, Zhang Y N.Discrete-time neural network for online timevarying nonlinear optimization with application to manipulator motion generation[J].IEEE Transactions on Neural Networks and Learning Systems, 2015,7(26):1525-1531.

        [24]張凱良,楊麗,張鐵中.草莓收獲機(jī)器人采摘執(zhí)行機(jī)構(gòu)設(shè)計(jì)與試驗(yàn)[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2011,9(42):155-160.Zhang Kailiang,Yang Li, Zhang Tiezhong.Design and experiment of picking mechanism for strawberry harvesting robot [J].Transactions of the Chinese Society for Agricultural Machinery, 2011, 9(42): 155-160.(in Chinese with English abstract)

        [25] TanN.Calibrationforaccuracyimprovementofserialmanipulators based on compressed sensing[J].Electronics Letters, 2015, 11 (51): 820-822.

        [26] Luciano C, Giovanni M, Marco N, et al.A joint-angle estimation method for industrial manipulators using inertial sensors [J].IEEE/ASME Transactions on Mechatronics, 2015, 5(20): 2484-2494.

        [27] Hamid S, Luigi V, Mehdi K, et al.Task-space control of robot manipulators with null-space compliance [J].IEEE Transactions on Robotics, 2014, 2(30): 493-505.

        [28] Liu YC,Nikhil C.Gravity-compensation-drivenpositionregulation forroboticsystemsunderinput/outputdelays[J].IEEETransactions on Control Systems Technology, 2014, 3(22): 995-1003.

        [29] Kern J, Jamett M, Urrea C.Development of a neural controller applied in a 5-DOF robot redundant[J].IEEE Latin America Transactions, 2014, 2(12): 98-106.

        [30] Spencer B B, Aaron M D.Robust resonant frequency-based contact detection with applications in robotic reaching and grasping[J].IEEE/ASME Transactions on Mechatronics, 2014, 5 (19): 1552-1561.

        [31] Corrado G, Lo B, Fabio G.Real-time planner in the operational space for the automatic handling of kinematic constraints [J].IEEE Transactions on Automation Science and Engineering, 2014, 3(11): 730-738.

        Series actuator end integrated positioning analysis based-on multilayer perceptron neural network

        Hu Yanzhu, Li Leiyuan
        (College of Automation, Beijing University of Posts and Telecommunications, Beijing, 100876, China)

        Abstract:It is difficult to establish Jacobian matrix and determine the coordinate frames of links for non-standard actuator.A new analytical method to establish the Jacobian matrix and determine the coordinate frames for joints and links are proposed in this paper.The proposed method made the positioning analysis of end-effector easier in space.At the same time, it is necessary to prove the effectiveness of the proposed method theoretically and verify the localization and configuration capabilities through simulations.First of all, forward kinematics model was set up based on a non-standard five Degree Of Freedom(5-DOF)actuator.A frame transformation is performed from base coordinate to end-effector coordinate.The relation between two adjacent joints is defined by a homogenous pose matrix.Secondly, the necessary and sufficient conditions for comprehensive localization are derived.They can guide the actuator to perform various tasks, such as tracking, assembly and autonomous grasping.A 5-DOF actuator is considered here as an example and this holds good for any N-DOF.Thirdly, inverse kinematics solutions are obtained by using artificial Neural Network(NN)based on backpropagation Multi-Layer Perceptron(MLP, multilayer perceptron NN)and are not unique.A unique solution using nonlinear minimization optimization is found.A NN based on supervisory learning method including three inputs, twenty neurons and five outputs has been used.Excitation function tansigmoid and linear excitation function pureline are in hidden and outer layers respectively.In Cartesian coordinate space, NN is trained by means of Levenberg Marquardt(LM)algorithm.The training sets used are Denav Hartenberg(DH)parameters and Cartesian coordinates.The weights are updated continuously which reduces the Mean Square Error(MSE)gradually.When MSE reaches the threshold set up, NN training will be terminated.After training, the test sets are used to examine the capability of NN.Fourthly, there are two evaluation functions viz., localization and cost functions.The localization function is defined to evaluate the positioning property of end-effector.At the same time, in task space, it will check whether the actuator has reached the target point along the direction needed or not.The cost function is defined to evaluate the kinematics configuration.There is a great relevance between cost function and Jacobian matrix.Velocity mapping from each joint to the end-effector was described by Jacobian matrix.So the cost function could give expression for kinematic configuration.At the end, simulations and experiments are conducted.The settings include industrial computer UNO2184G, 5-DOF non-standard actuator, Windows 7, MATLAB2012a.Coordinate frames for each joint are established and D-H parameters are determined.Then relative pose matrix is obtained between each of the two adjacent joints.Initial end-effector pose is obtained following right multiplication rule.The end-effector space range is formed under each joint operation range.Then, simulation is performed using NN, obtained localization and cost functions.The following results are obtained.The rank of Jacobian matrix is equal to 5.Therefore, this actuator met necessary and sufficient conditions for comprehensive positioning.NN method for solving inverse kinematics has reduced the computational complexity compared to conventional method.There are 21 groups of solutions when positioning to(41.4, 89.0, 104.5).The optimal solution obtained is(21.61, 91.44, 135.52, 221.42, 0)according to localization function rule.The optimal solution obtained according to cost function rule is(21.61, 125.73, 108.42, 221.99.41, 0).NN accuracy is 89.9%(approximately)while conventional method is 87.5% .By approximate estimation, the errors for θ1,θ2,θ3,θ4and θ5are 3.7°, 3.1°, 3.5°, 3.3°and 4.5°respectively.NN used 1.2 seconds while conventional method completed in 0.9 seconds.Therefore, computation accuracy has improved by 20% and efficiency by 2.4%.If the system is linear, the conventional method is chosen when less demand in real-time.In contrast, if the system is nonlinear, new method proposed in this paper is chosen when more demand in real-time.The minimum value of localization function is 0.96.The maximum value of cost function is 4.0349×1014.These two parameters decide the comprehensive positioning and the kinematics configuration.From the results presented, it can be concluded that the non-standard actuator with MLP has better localization and optimal configuration.

        Keywords:mechanization; control; models; manipulator; neural networks; inverse kinematics

        作者簡(jiǎn)介:胡燕祝(1970-),教授,博士后,博士生導(dǎo)師,主要從事視覺測(cè)量與機(jī)器人方向。北京北京郵電大學(xué)自動(dòng)化學(xué)院,100876。Email:YZH@263.net

        基金項(xiàng)目:北京市計(jì)劃課題《軌道交通事故現(xiàn)場(chǎng)應(yīng)急處置裝備研制與示范應(yīng)用》(Z131100004513006)

        收稿日期:2015-07-25

        修訂日期:2015-11-13

        中圖分類號(hào):TP212

        文獻(xiàn)標(biāo)志碼:A

        文章編號(hào):1002-6819(2016)-01-0022-08

        doi:10.11975/j.issn.1002-6819.2016.01.003

        猜你喜歡
        機(jī)械臂控制模型
        一半模型
        重要模型『一線三等角』
        重尾非線性自回歸模型自加權(quán)M-估計(jì)的漸近分布
        基于智能車技術(shù)的圖書館書籍整理機(jī)器人
        機(jī)械臂平面運(yùn)動(dòng)控制與分析
        機(jī)械臂工作空間全局相對(duì)可操作度圖的構(gòu)建方法
        人機(jī)交互課程創(chuàng)新實(shí)驗(yàn)
        3D打印中的模型分割與打包
        社會(huì)轉(zhuǎn)型期行政權(quán)控制的路徑探索
        科技視界(2016年21期)2016-10-17 17:06:18
        會(huì)計(jì)預(yù)算控制現(xiàn)狀及方法
        亚洲va中文字幕无码久久不卡| 久久精品国产清自在天天线| 精品无码一区二区三区爱欲| 亚洲精品动漫免费二区| 天天狠天天添日日拍| 日本丰满熟妇hd| 宝贝把腿张开我要添你下边动态图 | 色多多a级毛片免费看| 欧美疯狂做受xxxxx高潮| 精品午夜久久网成年网| 欧美人妻日韩精品| 人人添人人澡人人澡人人人人| 欧美一级欧美一级在线播放| 色综合久久加勒比高清88| 99热高清亚洲无码| 日韩av在线免费观看不卡| 美腿丝袜日韩在线观看| 在线麻豆精东9制片厂av影现网| 国产精品扒开腿做爽爽爽视频| 欧美性猛交xxxx乱大交丰满| 级毛片无码av| 久久亚洲精品中文字幕蜜潮 | 久久久中日ab精品综合| 77777亚洲午夜久久多人| 国产成人精品三级在线影院| 亚洲精品白浆高清久久| 成人亚洲av网站在线看| 91精品人妻一区二区三区久久久| 亚洲人成网站色7799| 女人让男人桶爽30分钟| 国产精品亚洲А∨天堂免下载| 亚洲va成无码人在线观看| 熟女高潮av一区二区| 国产最新女主播福利在线观看| 久久亚洲欧美国产精品| 亚洲精品久久久久高潮| 亚洲人成影院在线高清| 中文字幕中文字幕人妻黑丝| 按摩师玩弄少妇到高潮av| 艳妇臀荡乳欲伦交换h在线观看| 久久国产精品二国产精品|