Sunteți pe pagina 1din 9

Journal of Manufacturing Systems

Volume 11/No. 4

A Method for Enhancing the Accuracy of CNC


Machine Tools for On-Machine Inspection
d. Mou and C. Richard Liu, Purdue University, West Lafayette, IN

Abstract machine. A good alternative is to determine the


Efforts have long been devoted to developing a inaccuracy of a measuring machine and compensate
rapid and accurate method for dimensional inspection for it. We have developed methods for modeling
in a computer-integrated manufacturing environment. and compensating machine tool errors and have
In this research, a general method for enhancing the achieved a ten-fold improvement in positioning
accuracy of on-machine inspection is discussed. The
objective is to develop a reference-part based error
accuracy. 1-2 Thus, we have created the possibility
modeling algorithm to identify the parameters of posi- of using machine tools as inspection machines with
tioning error for a machine with minimal measurement. the needed accuracy.
Quadratic and cubic error models based on rigid body Machine tool errors can be classified as quasi-
kinematics have been developed. The accuracy and static errors and dynamic errors. Dynamic errors
complexity of both error models have also been dis- result from machine tool vibration, tool chatter, and
cussed. Experimental results demonstrate the effec-
tiveness of the error modeling algorithm with a ten-fold spindle vibration. Quasi-static errors are composed
improvement. This error modeling technique can also of the geometric error of the machine, influence of
be applied for error compensation, thus reducing the static loading, and thermal effects on the machine
errors in material removal processes. structure. Static loading and thermal effects change
the accuracy of the machine tool by altering the
Keywords: Machine Tool Accuracy, On-Machine geometric error, l'3n Dynamic errors contribute
Inspection, Error Modeling, Error Estimation and Com-
pensation almost no effect to the on-machine measurement
process, while quasi-static errors deteriorate the
accuracy of on-machine measurement. Therefore,
in this study, quasi-static errors are the main error
sources that must be determined.
Introduction Two major approaches have been proposed to
The on-machine inspection approach extends the model the quasi-static errors--the workspace and
function of a machine to include an automatic way element approaches. In the workspace approach, the
to sense and compensate for errors during a resultant positioning errors are modeled as some
sequence of machining operations. The dimensions distortion of an ideal work space. 7-11 This approach
of a workpiece can be checked on the spot against is relatively simple and valid for most configura-
the design and manufacturing standards to ensure tions of the machine tool, but it disregards the
the finished product is within the designated toler- behavior of an individual positioning element. In
ance. When workpieces are measured, it is common the element approach, the imperfection of each
to interpret the observed errors as the dimensional positioning element is addressed. 2,s'8,~21n The
inaccuracy of the workpiece. This can be true only resultant positioning error is treated as a function of
if the measuring devices are perfect. Otherwise, the the distortions of each positioning element. The
inaccuracy of a measuring device will influence the element approach with rigid body kinematics is
magnitude of the measurement. It would be too commonly used for CMM and machine tool testing,
costly to build and maintain a perfect measuring calibration, and error compensation. A similar tech-

229
Journal of Manufacturing Systems
V o l u m e 11/No. 4

nique has also been applied in dynamic model


parameter estimation and calibration for rigid-link Xl
X 1'
serial manipulators. 17
Many kinematic reference standards and artifact Xl

standard parts such as the magnetic ball bar, master . ~ . . . I F " x 3'

disk, ball plate, multi-directed bar, step gauge, and


metrology pallet have been used to identify geomet-
ric parameters or to determine the volumetric error
of CMMs and machine tools. 2'~°'18-22 Most work- CS l
x r,/
r /
pieces do not have the same geometric shapes and X3 la /

dimensions as the reference standards. Therefore, I/b


the possible error motions of the tool or probe tip
along the workpiece contour can only be estimated
with a degree of uncertainty.
The approach proposed in this paper deals with Figure 1
Relative Position of T w o C o o r d i n a t e Systems
mathematic models and measurements closely on an Inaccurate Link
related to the real parts and therefore reduces the
uncertainty in error estimation and error compensa-
tion. The newly proposed idea uses the machine tool 1 - ~i ]ti ai + Aai
to measure a reference-part with accurately known
oti 1 - [~i bi + Abi
dimensions and shapes. Using the difference Si=~
between the measured dimensions and the actual - ]ti [~i 1 c i + Ac i
one, we can model the machine tool errors with
rigid body kinematics. By using the error model, we 0 0 0 1
can correct the measured data of other inspected
parts. In this way, we remove the inspection error
from a machine tool. This error modeling process where o~i, 13~, and ~/i are the angular distortions of
can be easily transformed into the error compensa- the ith link; a i, b i, and c~ are the dimensions of the ith
tion process. Therefore, the proposed method can link; Aai, Abi, and Ac~ are the dimensional errors of
be applied not only to correct measurement error, the itn link. All error terms in this shape transform
but to also reduce machining error. matrix are assumed position-independent and are
treated as constants. The joint transformation matrix
describes the relative motion of two coordinate
systems along an imperfect joint, as shown in
Theoretical Background Figure 2, and can be written as:
Multi-axis machines typically composed of a
sequence of elements or links connected by joints
provide either a rotational or translational motion. 1 - 0[,xi ~¢xi xi + Axi
With rigid body kinematics, each element and joint
can be modeled as a homogeneous coordinate 0txi 1 - [~Xi Yi + Ayi
Ji =
transformation, z3 In real systems, some small
- 7xi 13xi 1 zi + Azi
imperfections exist in machine tool structural ele-
ments. To include those imperfections, the transfor- 0 0 0 1
mation matrix must be modified and can be decom-
posed into the shape and the joint transformation
matrices. 24 The shape transformation matrix where OLxi, fSxi, and ~/xi are the angular errors (roll,
describes the relative position of two coordinate pitch, and yaw) along the i th joint; x i, Yl, and zi are
systems on an inaccurate link, as shown in Figure 1, the desired displacements along the i joint; Axi,
and can be written as: Ay i, and Azi are the displacement errors along the :h

230
Journal of Manufacturing Systems
Volume 11/No. 4

xI
, x1

~ t ~ X 3'

x3 J / Moving
x2~1¢ ~. Axis
x 2'

Figure 2
Relative Motion of Two Coordinate Systems Along an
Inaccurate Joint

joint. All error terms in the joint transform matrix Figure 3


Structural Configuration of a Machining Center
are treated as functions of position and must be
specified to derive the specific error model for a
particular machine. For angular errors on axis x~:

The general homogeneous transformation matrix,


A/, for an inaccurate positioning element can be z~x2,(1) = f ~x, d x , ; Ax3,,,, - f [3x,dxl
expressed as A i = S i * Ji. This homogeneous
transformation matrix can be used to build a math- For angular errors on axis xz:
ematical model, describing the resultant positioning
errors, for any multi-axis machine with an arbitrary ~kXl,(2) -- f O~xzd~2 ; ~kX3,(2) = f "~xz dx 2
combination of rotary and translational joints. For
an n-axis machine, the relationship between the For angular errors on axis X3"
actual position and the desired position of the tool or
probe tip can be expressed as: Ax,,o)= fJ3x3dxa;Ax2,(3)- f3%dx3

[?actual] = [T] [?desired] where Axi,03 is the displacement error in axis i


where [T] = [A1] [A2] [A3] ... [An] resulting from angular errors on axis j.
The kinematic chain between the tool and the part
is closed by letting the origin of the tool coordinate
Error Estimation system be identical to the origin of the workpiece
and Compensation coordinate system. The relationship between the
To demonstrate the general procedure for error position vector of a point in the tool coordinate
estimation and compensation, the configuration of a system and the workpiece coordinate system can be
Cincinnati Milacron machining center (T-10) has expressed as:
been used to formulate the proper T matrix. For this [?workpiece] = [51] [ A ] [52] [J2] [83] [J3] [84]
particular machine tool configuration, as shown in
Figure 3, each axis is designed to have translational
[?,ool]
movement in only one direction. The displacement The positioning error vector is derived from the
error along the design movement direction is difference between the actual positioning vector in
approximated as a linear function of position. Any the workpiece coordinate system and the desired
displacement error in the other two constrained positioning vector in the tool coordinate system.
directions results mainly from the effect of angular [ep] = [?workpiece] -- [?tool] = [T] [?toot] - [?toot]
errors along the axis and is proportional to the joint -- [T-I] [?tool]
displacement.

231
Journal of Manufacturing Systems
Volume ll/No. 4

To solve the coefficients of the error model, we Cubic Form Error Model
must define the characteristics of the angular errors for Error Estimation (3-Axes)
in the joint transform matrix. To shed some light on Next, we approximate the angular errors along
the trade off between estimation accuracy and the axes as quadratic functions of position and
complexity, we will approximate the angular errors express them as follows:
along each axis as linear and quadratic functions of
position, respectively.
O£xi: Xi O~i, 1 + Xi2Oti,2 ~ x i : Xi ~i,1 + Xi2 ~i,2
"Yx, : Xi "~i,1 q- Xi2 "Yi,2
Quadratic Form Error Model By substituting the angular error terms into the Ji
for Error Estimation (3-Axes) and following the same procedure for matrix manip-
First, we assume that the angular errors along the ulation, the error vector of any point within the
axes can be approximated as a linear function of workspace can now be approximated as a cubic
position and expressed as follows: function of position.
Olxi = Xi Oti,1 ~x, = Xi ~i,1 ~x i = Xi "Yi,l
By substituting the angular error terms into Ji,
ep~Co+Cl ×'~ +C2 ×2x3 +C3 x22 +(24 x22xl +C5 x23
multiplying the matrices, rearranging the error X LXl X3 ~ X22 x3
X3- X32 Xl X33
terms, and ignoring the effect of second and higher
order error terms, the error vector of any point X32 X2

within the workspace, with respect to the common where


origin of two coordinate systems, can be approxi-
mated as a quadratic function of position.

~p=CO+C1
[x,] [x,x1
x2 + C 2
X3
X2X3 + C 3
Xl X3
x2 2
C3 = C3 +
0
0~i2
-~i2
- ~i2 ~/i2
0
~i2
/
- ~l.~_] £b,]
X32
where _ 20q2 2712 0 2~22 0 - O~32

C4= 0 - 2~12 1122 - 213= ct32 0

o -~cti Y-.7, 2~,12 0 - 722 0 - 732 [~32


I>l '=' '='
ep = 0 " ~ 1122 "/32+~ [~12
el: - i=l
C5 = ~ Ctl 2 0 - ~32" ~"/32
" o
- Vl2-313[h2 ~22 + ~ "~22
J
0

o - czii ~/i~
Ct = ~¢q+2cq
,~: ~dAx~ -2~-~ + O~il o " [~il Using the error models, the positioning error
-~/il 13. 0 vector at any point along the reference-part contour
" ~'~/i-2"/1 ~[]'i+2~i dAx'~
,*z d~x3
can be obtained. For the quadratic error model, 18
coefficients must be solved. On the other hand,
I - 2~11 2yil 2y21- ~313
there are 27 coefficient terms in the cubic error
C2 = a21 asl"2~n -2~21 model that must be solved. Obviously, the complex-
2~tl-y2t -Y3t ~sl
ity of the error modeling process increases with the
I 0 - 232~21 order of the approximation function. To obtain a
C3 = }all 0 - [~3t- 1 unique solution for the coefficients, an equal num-
0
ber of independent equations are needed. The error
-Yl1-½[31l ~321+ ~Y2I
model can be rearranged and formulated as

232
Journal of Manufacturing Systems
Volume 11/I'4o. 4

Cx -- e will screen all the feasible sets of measuring points


where x is an n x 1 coefficient vector, e is an n x and identify the sub-optimal set of measuring points
1 error vector, and C is an n x n matrix. The C once the strategic measuring points are selected
matrix is a function of positions and machine tool from the design specification. The search algorithm
constants. Based on the coordinates of the potential for the sub-optimal set of measuring points has two
measuring point selected, the property of the C criteria. First, the estimation residuals of the
matrix will vary. A unique solution for x can not be selected set of measuring points must be less than
found unless C "~ exists. It is difficult and time the allowable tolerance threshold designated by the
consuming to determine the proper set of potential user. Second, the selected set has the least sum of
measuring points that constitute a nonsingular C square of residuals.
matrix. The number of strategic measuring points is
To increase the effectiveness of the error model- determined by the user depending on the error
ing process, an iterative search algorithm, using the characteristics of the machine and the complexity of
recursive computational method, has been devel- the part's shape. The error modeling accuracy may
oped. The search algorithm, as shown in Figure 4, be improved if more measuring points are available
for selection. However, more measurements result
in more experimental and computational efforts,
which may not be desirable and practical. The user
MSelect Strategic"~
easuring Poin~J selects the strategic measuring points based on
experience and the inspection specification. The
positioning errors at those strategic measuring
points are then determined. The coefficients can
then be derived based on the measured positioning
errors for each combination of measuring points and
will be used to compute the estimated errors at each
measuring point. If the difference between the
estimated and measured error at any point is larger
than the pre-set threshold, the current combination
of strategic measuring points will be discarded.
Otherwise, the sum of squares of the residual in
each axis is recorded. The search procedure repeats
until all possible combinations are exhausted.
After all iterations have been completed, the set
of measuring points with the least resultant sums of
square will be identified as the sub-optimal set.
Once the sub-optimal set of measurement points is
determined, the measured errors at those points can
be used to solve the coefficients of the error model
with simple numerical methods such as Gauss
~ No elimination. The error model can then be used to
estimate the positioning error of the tool or probe tip
_ s-
along the reference-part contour and to correct the
(.o ustsot) measured data of on-machine inspection. The error
V estimation results cannot only be used to enhance
oo.o I the accuracy of on-machine inspection, but also to
reduce machining error by driving the probe or tool
tip more accurately along the desired part contour.
Figure 4 To accomplish this task, an appropriate compensa-
Iterative Search Algorithm tion signal must be determined and added to the

233
Journal of Manufacturing Systems
Volume ll/No. 4

original input signal to compensate for the position- Cubic Form Error Model for
ing errors. The relationship between the error com- Error Estimation (2-Axes)
pensation signal and the position vector of the tool
tip can be expressed as: exk =
i=l i=i i=l i=2 i=2

[-f;ctual] = [T] [Fdesired + -ec] + x2 Ic3 ~1- b3 a.~l - a~']- 2all xlx_,+ xl2 [ c3 YL2-~'.b3 a121
3 ~ z~=2 3 i=2
Rearranging the matrices, the error compensation + x2 :~[c3 y22- b3 a , ~ - ~- ct, I] - 2 a l ~ x l - x , - ~ - a ~ x, ~

signal can be obtained as:


ej~2 ~..Abi + a3 ~ c t t + a2 o - - c3 ~l~ - c2 [31 + xl [ ~ a i 0-11 - ~ c i I$1t + ¢t~]
i=l i=l i=l i=2 i=2
[ec] = [ T]-I IF*actual] -- [Fdesired] dAx, 2 3 3 3
.
+ X2 [a30'~ 1" c3 ~21 + ~ : x 2"] + ~21xlx2+ Xl [~.a3
i_~ 0~12- ~i=~
c 3 ~12 + ~ t i i]
= [ r -1 - - I ] [Fdesired]
2 ~ 4 3
. . . . . . . . . . . a *-
Using the same coefficients acquired from the
error estimation process, the inversion of the T where
matrix can be obtained numerically with the Gauss-
a~ = oL3 + 2oL z + 2 o q &
Jordon elimination method. The error compensation
signal at any point within the machine tool work- az = a3 + az + 2oL 1
space can then be calculated as a function of The experiment was conducted by using a Ren-
position. Adding this compensation signal to the NC ishaw touch-trigger probe (MP-3) to measure a
program for the subsequent machining operation round reference-part, as shown in Figure 5. The
will result in a significant improvement in the
dimensional accuracy of the finished part.

Experimental Verification
The experimental verification was conducted for
a two-dimensional model. From the transformation
matrices for inaccurate links and joints, the kine-
matic equation for this two-dimensional model is
written as:

[Fworkpiece] -~ [511 [J1] [52] [']21 [53] [F,ool]


Manipulating the resultant transformation matrix
in a similar manner, the expression for the position-
ing errors in the x 1 and x 2 direction can be obtained [ .. Pu/_. .
P0, L ~- P13 .~ Fixturing
as follows:
Testing
Point
Quadratic Form Error Model
for Error Estimation (2-Axes)
_ dAXll
ex, = ~Aai-b3~'~oq-b2. o q + c 3 ~ Y i + c 2 T l + x l [ ~ c i ~ l l ' ~ b i C q l ~ - x l 1
i=l i=l i=l i=2 i=2
+ X2 [C3 T21- b3 a 2 l - i f ' l ] - 2 a l l x l x 2 - ~ 0~21 x22

ex: = bi+a3Z~i+a2eq-c3Z~i'c2~'+xt[ alatl- ci[3H+~]


i=l i=l i=l i=2 i=2 Touch-trigger
dAx2] + 3 2 Probe ~ xl
+ g2 [a3°~21" c3 [~21 + dx 2 (221XlX2 + 2 O t l l x l

Figure 5
Reference-Part and Probing Configuration

234
Journal of Manufacturing Systems
Volume 11/No. 4

reference-part was mounted in the x l - x 2 plane that is


Measured v.s. Estimated Error along x2 Axis
perpendicular to the spindle where the probe is (Cold Condition)
inserted. It is assumed that no positioning error is 0.003
contributed from the x3 direction. Point Po was used
as a reference point for measurement calibration. 0•002 -
• " "4.

The observed errors at point Po characterize the drift


0.001 -
of the workspace. The errors at other measurement
points are considered relative to point Po and have 4),000 :
no constant term once Po is calibrated• The differ-
ence between the measured and referenced coordi- 4).001 -
nates was defined as error and used to calculate the
coefficients of the error model. Errors at several test -0.002

points were also measured and compared with the • Meas. Error
-0,003 ~ ..... ~ ..... Quad. Approx.
errors predicted by the error model. The measure- - --o- - Cubic Approx.
ment was conducted for a cold machine first• The 4).004 i i

machine was then warmed up for four hours by 9 13

moving the Xl and x 2 axes continuously at 50 inches Testing Point


per minute and rotating the spindle at 4000 RPM.
The same measurement was conducted again at
Figure 7
warm machine condition• Measured and Estimated Errors Along x2
The measured data and the results of quadratic Axis for Cold Condition
and cubic approximations are shown in F i g u r e s
6-9. The data show that the positioning error along the x 2 axis. The results also show that, in all cases,
the x~ and x z axes has the same order of magnitude the cubic approximation has more accurate results,
at cold condition. As the machine warms up during though more measurement efforts are involved. The
the experiment, the positioning error along the x 1 mean and the sum of square (Zexi2) of the residuals
axis has a much larger change in magnitude than in for quadratic and cubic approximations along both

Measured v.s. Estimated Error along xl Axis Measured v.s. Estimated Error along xl Axis
(Cold Condition) (Warm Condition)

0.003 0.003 [

0.0021 0.002 t

0.001 1

4).000 '

4).001 4).001

-0.002 -O.002
• Meas. Error
-0.003 ' ...... n- .... Quad. Approx. -0.003 .....M/ . . . . . n. .... Quad. Approx
---o-- Cubic Approx. ~ - o--. CubicAppr 7 .
-0.004 i
-0.004
9 13 5 9 13

Testing Point Testing Point

Figure 6 Figure 8
Measured and Estimated Errors Along x 1 Measured and Estimated Errors Along x~
Axis for Cold Condition A x i s for Warm Condition

235
Journal of Manufacturing Systems
Volume 11/No. 4

Table 2
Measured v.s. Estimated Error along x2 Axis Sum of Square of Residuals for Cubic and Quadratic
(Warm Condition) Approximations
0.003

Quadratic Cubic
0.002 Approximation Approximation

0.001 Z~l Zex22 Ee21 Ze~

-0.000
Cold
Condition 3.14 * 10 .6 1.17 * 10 .6 8.10 * 10 -7 9.00 * 10 .8
-0.001
t~

-0.002 Warm
Condition 3.94 * 10 -6 8.30 * 10-7 6.50* 10 -7 4.60* 10 -7
• Meas. Error
-0.003 ...... n ...... Quad. Approx.
- - -o - - Cubic Approx.
-0.004 i
9 13
the cubic approximation is approximately 6 and 2
Testing Point
times smaller than that of the quadratic approxima-
tion. The results also show that the cubic approxi-
Figure 9 mation has a more distinct performance than the
Measured and Estimated Errors Along x2
Axis for Warm Condition quadratic approximation when the positioning error
has a large range of change.
axes are shown in Tables 1 and 2. For cold
condition, the mean of the residuals for the cubic
approximation along the x 1 and x 2 axes is approxi- Conclusion
mately 1.93 and 2.73 times smaller than that of the In this study, a general method for enhancing the
quadratic approximation. For warm condition, the accuracy of CNC machine tools for on-machine
mean of the residuals for the cubic approximation inspection has been developed. With the difference
along the x 1 and x 2 axes is approximately 2.70 and between the measured and the referred dimensions
1.41 times smaller than that of the quadratic approx- of the reference-parts, the machine tool errors can
imation. For cold condition, the Eexl 2 and Eex22 for be modeled. By correcting the measured data with
the cubic approximation is approximately 4 and 13 predicted errors, the machine tool can be used as a
times smaller than that of the quadratic approxima- measurement machine up to the accuracy of the
tion. For warm condition, the Eexl z and Ee,,2 z for error modeling process. Many significant benefits
can be gained from this approach. For example, the
Table 1 workload and setup time for CMMs and the rework
Mean of Residuals for Cubic and Quadratic Approximations cost can be reduced due to the higher accuracy in
measurement and machining.
Quadratic Cubic Experimental results show that the error models
Approximation Approximation
can estimate the machine tool errors with desirable
exl ex2 exl ex2 accuracy. The results also show that, for this
particular application, cubic approximation yields
Cold
better results than quadratic approximation, though
Condition 3 . 5 0 " 10 -4 2.0625 * 10-4 1.8125 * 10-4 5 . 6 2 5 " 10 -5
more measurement efforts are involved. For other
applications, the same modeling technique can be
used to derive an error model corresponding to the
Warm
Condition
3 . 8 7 5 " 10 -4 1.9375 * 10 .4 1.4375 * 10 -4 1 . 3 7 5 " 10-4 error characteristics of that particular machine tool.
The user can start with a lower order error model
and gradually increase the order of the error model

236
Journal of Manufacturing Systems
Volume 11/No. 4

if the modeling result is not satisfactory. For imple- 15. G. Zhang, R. Ouyang, B. Lu, R. Hocken, R. Veale, and A.
Donmez, " A Displacement Method for Machine Geometric Error of
mentation, the machine tool can be calibrated by Machining Center," CIRP Annals, Vol. 37, 1988, pp. 515-18.
measuring the reference-part at cold machine con- 16. B.V. Kreng, C.R. Liu, and C.N. Chu, " A Compact Three-
Dimension Error Model of Machine Tools," Control Issues in
dition. As the cutting processes proceed, the effect Manufacturing Processes, DSC-Vol. 18, ASME Winter Annual
of thermally induced errors can be estimated and Meeting, San Francisco, CA, December 10-15, 1989, pp. 25-31.
17. M.E. Sklar, J.C. Hudgens, and D. Tesar, "Dynamic Model
compensated for by periodically recalibrating the Calibration for Rigid-Link Serial Manipulators," Computers in Engi-
machine tool. More reports on thermal effect track- neering, ASME International Computers in Engineering Conference,
Boston, MA, August 5-9, 1990, pp. 101-09.
ing will be presented in the near future. 18. J.B. Bryan, " A Simple Method for Testing Measuring Machines
and Machine Tools, Part 1: Principles and Applications," Precision
Engineering, Vol. 4, 1982, pp. 61-69.
19. F. Jouy and I.S.M.C.M., "Theoretical Modelisation and Exper-
References imental Identification of the Geometrical Parameters of Coordinate-
Machines by Measuring a Multi-Directed Bar," CIRP Annals, Voi.
1. M.A. Donmez, D.S. Blomquist, R.J. Hocken, C.R. Liu, and 35, 1986, pp. 393-96.
M.M. Barash, " A General Methodology for Machine Tool Accuracy 20. W. Knapp, "Circular Test for Three Coordinate Measuring
Enhancement by Error Compensation," Precision Engineering, Vol. Machines and Machine Tools," Precision Engineering, Vol. 5, 1983,
8, 1986, pp. 187-95. pp. 115-24.
2. P.M. Ferreira and C.R. Liu, " A Contribution to the Analysis and 21. H. Kunzmann and F. Waldele, "On Testing Coordinate Measur-
Compensation of the Geometric Error of a Machining Center," CIRP ing Machines (CMM) with Kinematic Reference Standards (KRS),"
Annals, Vol. 35, 1986, pp. 259-62. CIRP Annals, Vol. 32, 1983, pp. 465-68.
3. J.B. Bryan, "International Status of Thermal Error Research," 22. K. Busch, H. Kunzmann, and F. Waldele, "Calibration of
CIRP Annals, Vol. 16, 1968, pp. 203-15. Coordinate Measuring Machines," Precision Engineering, Vol. 7,
4. G. Spur and P. DeHaas, "Thermal Behaviour of NC Machine 1985, pp. 139-44.
Tools," M.T.D.R. Conference Proceedings, Vol. 14, 1973, pp. 23. J. Denavit and R.S. Hartenberg, " A Kinematic Notation for
267-73. Lower-Pair Mechanisms Based on Matrices," Journal of Applied
5. R.J. Hocken, "Technology of Machine Tools," Machine Tool Mechanics, 1955, pp. 215-21.
Accuracy, Vol. 5, 1980, UCRL-52960-5, Lawrence Livermore Lab- 24. P.M. Ferreira and C.R. Liu, "An Analytical Quadratic Model for
oratory, University of California, Livermore, CA. the Geometric Error of a Machine Tool," Journal of Manufacturing
6. M. Weck and M. Schmidt, " A New Method for Determining Systems, Vol. 5, No.I, 1986, pp. 51-63.
Geometric Accuracy in the Axis of Movement of Machine Tools,"
Precision Engineering, Vol. 8, 1986, pp. 97-103.
7. D.L. Leete, "Automatic Compensation of Alignment Errors in
Machine Tools," International Journal of Machine Tool Design
Research, Vol. 1, 1961, pp. 293-324.
8. W.J. Love and A.J. Scarr, "The Determination of the Volumetric
Accuracy of Multi-Axis Machines," M.T.D.R. Conference Proceed- Authors' Biographies
ings, Vol. 14, 1973, pp. 307-15. Jong-I Mou received his BS in mechanical engineering in 1986
9. R.J. Hocken, A. Simpson, B. Borchardt, J. Lazar, C, Reeve, and and MS in manufacturing system engineering in 1988 from the
P. Stein, "Three Dimensional Metrology," CIRP Annals, Vol. 26, University of Wisconsin-Madison. He is now working toward a
1977, pp. 403-408. PhD in the School of Industrial Engineering at Purdue University.
10. S.K. Lee, "Accuracy Improvement of a CNC Machining Center His research interests include precision engineering, intelligent
by Using a Touch-Trigger Probe and a Metrology Pallet," PhD machinery, manufacturing processes control, and computer-aided
dissertation, Purdue University, West Lafayette, IN, 1986. manufacturing. He is a member of Tau Beta Pi, Alpha Pi Mu,
11. H.K. Jan, C.N. Chu, and C.R. Liu, " A Configuration Indepen- Omega Rho, ASME, SME, CASA/SME, and RI/SME.
dent Approach for Modeling Three Dimensional Error Map for
Machine Tools Using Hyperpatch Model and Metrology Plate," C. Richard Liu received his PhD from Purdue University in 1972.
Modeling of Machine Tools: Accuracy, Dynamics, and Controls, His experience includes positions as a research engineer at Whirlpool
PED-Vol 45, ASME Winter Annual Meeting, Dallas, TX, November Corporation and as an assistant professor of mechanical engineering at
25-30, 1990, pp.75-85 Stanford University. He has been affiliated with Purdue University
12. R. Schultschik, "The Components of Volumetric Accuracy," since 1978 and is now a full professor of Industrial Engineering. Dr.
CIRP Annals, Vol. 25, 1977, pp. 223-27. Liu's current research activities include various aspects of computer-
13. J. Tulsty, "Testing of Accuracy of Machine Tools," UCRL- aided design and manufacturing such as conceptual design, precision
52960-Supp. 1, Lawrence Livermore Laboratory, University of Cal- engineering, and integrated planning.
ifornia, Livermore, CA, 1980. Dr. Liu was nominated for the Whirlpool Sigma XI Research
14. N.A. Duffle and S.J. Malmberg, "Error Diagnosis and Compen- award in 1975. He was a recipient of the 1981 Outstanding Young
sation Using Kinematic Model and Position Error Data," CIRP Manufacturing Engineer award of SME, the 1984 Blackall Machine
Annals, Vol. 36, 1987, pp. 355-58. Tool and Gage award of ASME, and the 1985 IR 100 award.

237

S-ar putea să vă placă și