您好,欢迎来到华佗小知识。
搜索
您的当前位置:首页Response Sensitivity Analysis of the Dynamic Milling Process Based on the Numerical Integration

Response Sensitivity Analysis of the Dynamic Milling Process Based on the Numerical Integration

来源:华佗小知识
CHINESE JOURNAL oF MECHANICAL ENGINEERrNG 940・ Vo1.25,No.5,2012 jI:10.3901/CJME.2012 05.940.available online at www.springerlink.com;www.cjmenet.com;www.cjmenet com cn Response Sensitivity Analysis of the Dynamic Milling Process Based on the Numerical Integration Method DING Ye ,ZHU Limin ,ZHANG Xiaojian ,and DING Han , 1 State Key Laboratory ofMechanical System and Vibration,Shanghai Jiao Tong University,Shanghai 200240,China 2StateKeyLaboratory ofDigitalManufacturingEquipmentandTechnology, Huazhong University ofScience and Technology,Wuhan 430074,China Received January 12,2011;revisedMay 11,20l1;accepted June 2,2012 Abstract:As one of the bases of gradient—based optimization algorithms,sensitivity analysis 1s usually required to calculate the derivatives of the system response with respect to the machining parameters.The most widely used approaches for sensitivity analysis are based on time—consuming numerical methods,such as finite diierence methods.Thifs Paper presents a semi—analytical method for calculation of the sensitivity of the stability boundary in milling.After transforming the delay—differential equation with time—periodic coeficientsf governing the dynamic milling process into the integral form,the Floquet transition matrix is constructed by using the numerical integration method.Then the analytical expressions of derivatives of the Floquet transition matrix with respect to the machining parameters are obtained.Thereafter,the classical analytical expression of the sensitivity of matrix eigenvalues is employed to calculate the sensitiviy of tthe stability lobe diagram.The two—degree—of-freedom milling example illustrates the accuracy and eficiency fof the proposed method.Compared with the existing methods,the unique merit of the proposed method is that it can be used for analytically computing the sensitiviy of tthe stabiliy boundarty in milling,without employing any finite difference methods.Therefore, the high accuracy and high eficiency are botfh achieved.The proposed method can serve as an effective tool for machining parameter optimization and uncertainty analysis in high—speed milling. Key words:milling,stability,sensitivity ofthe stability boundary,numerical integration method Altintas and co.workers proposed the zero order solutionL . 1 Introduction the multi.frequency solutionl。 .and an improved time domain method for low radial immersion milling J. Nowadays,the high speed milling technology is widely INSPERGER.et a1【 I Ipresented the semi—discretization used in industry to fulfil1 the rapidly growing demand for methods.BAYLY,et all and MANN.et al【”J developed ,manufacturing meta1 products accurately and eficiently.As fthe fundamentalissues in the research of high speed milling, et al[14】and YIet al[15]explored the Lambert W function chatter stability analysis and dynamics—based machining based method.BUTCHER.et alI 。 71proposed the parameter optimization have been extensively investigated Chebyshev polynomial based method and Chebyshev in the past two decades.As for the former issue,recent collocation method.More recently,DING et al【 …..the temporal finite element analysis(TFEA)method.ASL, ,, overviews in this field can be found in Refs.[1—3].In presented the fu11.discretization method and the numerica1 genera1.based on the experimental modal test technique of cutter.spindle systems and the cutting force model the dynamic milling process taking the regenerative effect into integration method. As or fdynamics—based machining parameter optimization.much effort has also been devoted.KURDI, al[20J.account is formulated as a delay—differential equation(DDE) et with time.periodic coefficients.The main task of chatter stability analysis is to predict the stability 1obe diagram of discussed the problem of simultaneous optimization of removal rate and surface 1ocation error on the basis of TFEA method.BUDAK.et a1 .proposed a machining parameters,providing the basis of selecting the optimal machining parameters.There exist many articles ocusing on tfhis mode1.based stability prediction problem. method for the maximization of chatter—free material removal rate based on the zero order solution of milling stability.MERDOL.et al【 Jpresented an optimization .strategy considering different machining constraints such as Corresponding author E—mail:hding( sjtu edu.cn This project is supposed by National Key Basic Research Program(973 Program. Grant No. 2O 1 1 CB706804),Nationa1 Natural Science Foundation of China rGrant No.50805093).and Science&Technology Commission ofShanghai Municipality,China fGrantNo.09QH1401500) 0 Chinese Mechanical Engineering Society and Springer-Verlag Berlin Heidelberg 20I2 chip thickness,cutting forces,spindle torque power,system stability,and forlTl errors on the workpiece to optimize machining parameters.However,there has been very little attention to response sensitiviy analtysis of the milling CHINESE JOURNAL OF MECHANICAL ENGINEERING process,which serves as the foundation of gradient—based optimization algorithms.Kurdi,et al ,proposed an ・941・ Mq(t)+C (f)+Kq(t)= approach for computing the sensitivity of the stability ap O) ( )一q(t一 )]+日 f0(t), where l1) boundary,using the adioint method in combination with the inite diffference method. C g( ,and a are the modal mass,damping, In this Paper,we present a semi.analytical method for calculation of the sensitivity of the stability boundary. Within the framework of the numerical integration method. the analytical expressions of derivative of the Floquet stiffness matrices,displacement vector,and axial depth of cut,respectively;T=60/(N ̄2)is the tooth passing period and N is the number of the cutter teeth;apfo(f) denotes the steady force item andf0(f)=f0O+ );Kc(f) denotes the coeficient matrifx indicating the component of transition matrix with respect to the machining parameters are obtained.On this basis.the classical analytica1 expression of the sensitivity of matrix eigenvalues[24J is cutting forces.Kc f )is presented as utilized to calculate the sensitiviyt of the stability boundary. One of the advantages of the proposed method is that it can be used for analytically computing the sensitiviyt of the stability boundary,without employing any finite difference methods.The rest of the paper is organized as follows. Section 2 gives the mathematical mode1 of the milling dynamics and briefly introduces the numerical integration method.The proposed method for calculation of the sensitiviyt of the stabiliyt boundary is presented in section 3.Simulation examples are given in section 4.Finally, Section 5 concludes the PaDer. 2 Mathematical Model and the Numerical Integration Method Within the framework of structural dynamics and using the experimental modal testing technique,the milling process is generally modeled as a two degrees of freedom (DOF)lumped—parameter dynamic system as shown in Fig.1,where卉( denotes the angular position of the cutter tooth(,), denotes the spindle speed,aP represents the radial depth of cut,[is the feed direction of the workpiece relative to the cutter,Ft |and Fr |are the tangential and normal cutting force components for the cutter tooth(,), respectively. Fig.1.Schematic representation of the two DOF milling process Based on the linear cutting force model for and 4]’the dynamic milling process taking the regenerative r.,[effect into account can be formulated as the following DDE[13,251: Kc(t)= 一h (t) ((ff))  (2) 一 (f) where Ⅳ hxx(t)=∑g( (f))sin( (f)× =1 [ c。s( o)+ sin( o)], (3) Ⅳ hxy(t)=∑g(q)j(t)cos(q)j(t)x =1 [ c。s(qJj(t)+Kn sin( (f)], (4) Ⅳ (f)=∑g( (f)sin( (f)× ,=1 sin( ̄j(t)+ cos( (f)J' N (f):∑g( ( )cos( (f)× =1 [一K,sin( )+ c。s( o)] .In Eqs.(3)一(6), and Kn are the tangential and the normal linearized cutting force coefifcients,respectively, and (f)is the angular position of the jth tooth defined by ∽:百2nS2 一 In Eqs.(3)一(6),the function g(力( )is utilized to indicate whether thejth tooth in or out of the cut,i…e ㈣ where礁t and are the start and exit angles of the jth tooth,respectively.Denoting by aP/D the radial immersion atio, as for the case of down—milling, t arccos(2a /D一1)and =it;as for the case of up。milling, t=0 and x=arccos(1—2a /D).Clearly, D1NG Ye,et al:Response Sensitivity Analysis ofthe Dynamic Milling Process Based ・942・ on the Numerical Integration Method Kc( )=K (f+ ). To discuss the chatter stability of the system.corresponding the static sampled time points are =t0+t,+( 一1) , where i=1,2,…,n+1 and force item ap To(f)can be dropped(while for the problem X(t1)=X(to+tj).We adopted Newton—Cotes formulas to of prediction of the surface 1ocation error,this item should calculate Eq.(1 1). be considered;see Refs.[13,26]).Then,Eq.(1)carl be At the discrete time points t (i=2,3,…,月+1),the reduced as corresponding response x(ti)can be gotten from Eq.(1 1) a Mq(t)WC!l(t) ̄Kq(t)=al: (,f) (f)一q(t一 )].(9) x(ti)=exp(A(tf—tf—1)) ( 一1)+ The key task of the model—based chatter stability analysis is to determine the stability lobe diagram of the spindle speeds and depth of cuts according to Eq.r91.For the sake of completeness,the trapezoidal rule based numerical {eXp( 一 ) )一 一 )] ,By using the trapezoidal rule[ Eq.(1 3)is simpliifed integration method【 J for prediction of chatter stability is as briefly summarized below.Denoting )+ )=eXp( 一 )) _1)+ap{ [eXp( 一 ))× (ti-1)( (f卜 )一 ( -1-T))+曰( )( ( )一 ( 一 ))】1. (14) ng Eq.(91 is re—presented in the following state—space form as 【,)=Ax(t)+a (f)【 【,)一x(t一 )], where A=1I M一 \一M_、Ci2  .2 CM—C/4一K CM /。 I-G~.aptD 2 x(q) : ● aptx(t ~7_1 D+E n(t1= o o1 x(t ) x(t +1一T) (15) (f)o J Denote by to,t,and tc the time the cutter leaves the workpiece,the duration of the free vibration(the time spent where out of cut)and the duration of the forced vibration(the time spent in cut).respectively.Denoting by P the ratio of to we can get tc=60p/(NX2、 and t,=60(1一P)/(Ⅳ ). 0 exp(Ar)0 G Considering the item ap )l ( )一 (f一 )I in Eq.(10) as the nonhomogeneous term of the homogeneous equation exp(Ar)0 exp(Ar)0 (f)=Ax(t),the solution of Eq.(1 0)is written as the ollowifng integral form 0 x(t)=exp(A(t—t0)) (fo)+ exp(Ar)B1 B2 {exp( ) ) )一 一 ) ,…)。= exp(Ar)B2 exp(Ar)B Bn+ where X(to)denotes the state value at t=tn.When the cutter is out of the cut,the term ( )in Eq.(1 1) vanishes, and Eq. (11) becomes E= 0…0 0 0 0 0 exp(Ar) 0 0 x(t)=exp(A(t—t0)) ( 0).At the end ofthe free vibration duration,i.e.,at the time t=to+tr,the system state becomes X(to+tj)=exp(AQ) ( ) And B denotes B(t )for i=1,2,・一,,z+1(note that To deal with the forced vibration process,the time the list of B(t )is independent of the spindle speed for a interval[to+ ,,to+ ]is equally divided into small fixed radial immersion ratio口 /D). intevarls.and the time step is denoted by f=tc/n.The Denoting that CHINESE JOURNAL OF MECHANICAL ENGINEERING 2 .F= I—G一.ap'C D = ・943・ apZ"D口 +E m the Floquet transition matrix on one tooth passing period can be expressed as dr d Aexp(Ar)0 Aexp(Ar)D Aexp(Ar)0 ..................唧m : , I 一 J1 m (16) In Eq(18),the derivatives of the duration of the 一2 .口口ransittion matrix are tess than one for a given axial depth m of cut ap and spindle speed Q tm he milling process is stable. 3 Response Sensitivity Anal州 ysis 3.1 Derivative of the Floquet transition matrix with respect to the machining 眦 parameters Firstly,by using Eq.(1 6),the derivative of the Floquet transition matrix with respect to(w.r.t.)the axial depth of cut p can be obtained as 丝: '菖 a口 三2 2。十三 Dtp-1F】J .Similarly,we can obtain th眦 e derivative of the Floquet transition matrix w.r.t.the spindle speed i2 as m e一 : r0 f1 一  0-(2 0 甜 n甜 ], (18) m where 0F D——..aptdr ......—— ・——× a d l A exp(Ar)B1 0 A exp(Ar)B2 0 + A exp(Ar)B exp(At,)A df, d Og"一0 【『 2一  d2 ̄』l D一一ap2 tdf : ——× d 口 A exp(Ar)B1 0 A exp(Ar)B2 0 A exp(Ar)Bn 0 df 60p d .QZNn (20) 3.2 Sensitivity of stability boundary After the derivative of hte transition matrix iS obtained. the classical analytical expression of the sensitivity of matrix eigenvalues[ 4j iS employed to calculate the sensitivity of stability boundary.For the sake of completeness,the next steps are given below by following【 . The derivative of an eigenvalue of the transition matrix W.r.t.a generic variable Z【 (in this paper,it can be hte axial depth of cut or the spindle speed1丝(l l{ 3 iS  一鱼 一 (21) where 1,and U are the left and right eigenvectors associated with the eigenvalue ,respectively,and the symbol(・,・) denotes the scalar product of two vectors.Then,the maximum modulus of eigenvalue I'31max w.r.t.the variable Z is '3max1_ ̄max 3'max0: — ,Z 2l l (22) where the symbol denotes the conjugate of the complex number . At last,the sensitiviyt of stabiliyt boundary at a given spindle speed and its corresponding critical axial depth of cut aplim can be obtained as ol'3 18Q = d p ap=aplim d ̄2 力=墙 口p=Ⅱpl蚰 ol'3 l。 (23) 力=日 ap=aplim DING Ye,et al:Response Sensitivity Analysis of the Dynamic Milling Process Based ・944・ on the Numerical Integration Method 4 Numerical Examples Computer programs of the proposed approach are all u』三implemented on a personal computer(Intel Core(TM)2 Duo Processor,2.1 GHz.1 GB1 by using MATLAB 7.4. The benchmark example in the literature is used to verify the validity and effectiveness of the proposed method. The dynamic equation of a two DOF milling model of Ref.[23】is used.The cutter modal parameters are given in Tab1e 1.The cutting conditions are:down milling D= 25.4 Inln,.Ⅳ=1, =0.1 mm/tooth,Kt=600 N/mm , 0 l2 13 l4 l5 16 17 1 8 l9 20 K =180 N/mm ,and the radial depth ofcut a =5 inln. The stability 1obe diagram via the numerical integration method is shown in Fig.2.The comparison of the sensitivity of the stability boundary by using the central ifnite difference method【 and the proposed method are both illustrated in Fig.3.The computation parameters for the two different methods are 1isted in 11able 2.where denotes the error tolerance for the bisection method iterations to determine the stability boundary,and h denotes the step size i— —c一 一 盖 £一n spi ndle speed for the centra1 f一 一i=一nite  一l一difference method .Fig.3 shows a good agreement of the two methods.Note t如 0 hat加 如  the proposed method provides a 如∞如 加 semi.analytical procedure to calculate the sensitivity of stability boundary without employing any finite difference methods.The high computational efifciency of the present method is achieved.The centra1 finite difference method takes about 24.3 s,while the proposed method only takes al:lout 8.3 s. Table 1.Cutter modal parameters from Ref.【23l Spindle speed n{(kr・rain- 、 Fig.2.Stability lobe diagram using the numerical integration method Spindle speed (kr・rain 1 (a)The central finite difference method c一∞ 加 如 O 11 l2 l 3 l4 l5 16 】7 1 8 19 20 Spindle speed (kr・rain 、 (b)The proposed method Fig.3.Comparation ofthe sensitivity ofthe stabiliyt boundary Table 2.Calculation parameters used in Fig.3 5 Conclusions f 1)In this paper,a semi—analytical method for calculation of the sensitivity of the stability boundary is proposed. Within the framework of the numerical integration method. the analytical expressions of derivative of the Floquet transition matrix with respect to the machining parameters are obtained.On this basis.the classical analytical expression of the sensitivity of matrix eigenvalues is utilized to calculate the sensitivitv of the stability boundary. f21 The main advantage of the proposed method is that it can be used ofr analytically computing the sensitiviyt of the stability boundary,without employing any finite difference methods.The accuracy and computational efifciency of the proposed method are both demonstrated. f31 The present work focuses on the topic of calculation  一∞ 如 ∞ 0 旧DING Ye,et al:Response Sensitivity Analysis of the Dynamic Milling Process Based ・946・ on the Numerical Integration Method for micro/nano manufacturing Huazhong Universiyt ofScience and Technology,China.in l989. SuppoSed by the Alexander von Humboldt Foundation,he E—mail:zhulm@sjtu.edu.cn WOrked at Universiyt ofStuttgart,Germany.from l993 to 1994. zHANG Xiaojian.born in 1 982.iS currently a PhD candidate at From 1994 to l996,he worked at School of Electrical and ectronic Engineering,Nanyang Technological University, State Key Laboratory of Di西tnl Manufacturing Equipment and Elngapore.From l997 to 2001.he was a professor at Huazhong Technology,Huazhong University of Science and Technology, SiChina.He received his BS and MS degrees in mechanical Universiyt ofScience and Technology,China.He ioined Shanghai engineering from Wuhan University of Technology,China,in 2004 and 2007.respectivelY.His research interests include machining dynamics and machining process optimization. Jiao Tong Universi China in September 2001.Prof.Ding was the recipient of Nationa1 Distinguished Youth Scientiic Fund of fChina in 1997.And he iS a senior member ofIEEE and a technical editor for the IEEE/ASME Transactions on Mechatronics.His research interests include robotics.multi.axis machining and equipment automation. E—mail:xizhang@rocketmail.com D1NG Han,born in 1 963,is currently a professor at Shanghai Jiao Tong University,China.He obtained his PhD degree from Tel:+86.2l一34206086;E.mail:hding@situ.edu.ca 

因篇幅问题不能全部显示,请点此查看更多更全内容

Copyright © 2019- huatuo0.cn 版权所有 湘ICP备2023017654号-2

违法及侵权请联系:TEL:199 18 7713 E-MAIL:2724546146@qq.com

本站由北京市万商天勤律师事务所王兴未律师提供法律服务