International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Contents lists available at ScienceDirect

International Journal of Machine Tools and Manufacture

journal homepage: www.elsevier.com/locate/ijmactool

Machining accuracy improvement of non-orthogonal five-axis machine tools

by a new iterative compensation methodology based on the relative motion
constraint equation
Changjun Wu a, b, Jinwei Fan a, *, Qiaohua Wang b, Dongju Chen a
a
College of Mechanical Engineering & Applied Electronics Technology, Beijing University of Technology, Beijing, 100124, PR China
b
School of Mechanical and Electrical Engineering, Zhengzhou University of Light Industry, Zhengzhou, 450002, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: This paper proposes a new iterative compensation methodology of geometric errors to improve the machining
Machining accuracy accuracy of a non-orthogonal five-axis machine tool (NOFAMT). Firstly, based on homogeneous transform matrix
Non-orthogonal five-axis machine tool (HTM) and multi-body system (MBS) theory, the relative motion constraint equations (TRMCEs) of the tool tip
Iterative compensation methodology position and tool orientation vector related to a NOFAMT with a nutating rotary B axis are established. Then, by
The relative motion constraint equation utilizing TRMCEs, the mapping relationships between tool path and the numerical control (NC) command without
Geometric error
and with considering the geometric errors are constructed respectively. In order to truly reproduce tool motion
trajectory of the machine tool driven by the given NC command, the mapping relationship between the NC
command and tool cutting trajectory is also established. Meanwhile, procedures of iterative compensation are
described by using the aforementioned mapping relationships without the traditional inverse calculation, and the
actual NC code is generated in self-developed compensation software. It is not difficult to find that the new
approach takes the difference between tool path and tool cutting trajectory as the control objective and can
directly obtain the actual NC code controlling the machine tool to achieve the desired machining accuracy.
Finally, a cutting test is carried out on the DMU60P NOFAMT. Experimental results show the developed iterative
compensation methodology is precise and effective for NOFAMTs. Therefore, compared with the existing
methods, the new method is more direct and accurate. And its basic idea can be applied to other type of machine
tools.

1. Introduction way [5].

With the increasing requirements for higher quality machined com-
plex parts, the importance of improving machine tools' machining ac-
curacy is well recognized in the manufacturing community [1].
Compared with three-axis machine tools, multi-axis machine tools with
one or more rotary axes have many advantages, such as higher
metal-removal rate, lower cutting time, higher production rate, fewer
set-ups [2,3,7–9], better versatility and flexibility, higher machining ef-
ficiency [4]. However, due to the additional rotary axes, multi-axis ma-
chine tools are introduced more error sources, thus bringing about lots of
barriers and difficulties in machining accuracy improvement. Among
them, geometric errors are a basic factor resulting in poor machining
accuracy [5,6]. Fortunately, the geometric errors can be reduced by error
compensation, which has been proven as an economical and effective
At present, considerable research work for error compensation of
multi-axis machine tools is devoted almost entirely to the orthogonal
series machine tools [2–7,9,12–18,23,25], whose three linear axes are
orthogonal each other and rotary axis center line is parallel to the di-
rection of the linear axis [10]. However, to the best of our knowledge,
less attention is paid to error compensation of non-orthogonal multi-axis
machine tools, whose rotational axis is in an inclined plane [11].
Unlike the orthogonal multi-axis machine tool, a non-orthogonal
multi-axis machine tool with a nutating head or a nutating table can
adjust the position and orientation of the cutting tool with respect to the
workpiece toward any angle within the workspace. In addition, it also
can travel continuously between the horizontal and vertical positions on
the same machine [10,11]. Hence, the research on error compensation of
non-orthogonal multi-axis machine tools is an important topic.

* Corresponding author. College of Mechanical Engineering & Applied Electronics Technology, Beijing University of Technology, Ping Leyuan 100#, Chaoyang district, Beijing, 100124,
PR China.
E-mail addresses: piaoyanggh2003@163.com (C. Wu), jwfan@bjut.edu.cn (J. Fan).

https://doi.org/10.1016/j.ijmachtools.2017.07.008
Received 6 May 2017; Received in revised form 27 July 2017; Accepted 28 July 2017
Available online 31 July 2017
0890-6955/© 2017 Elsevier Ltd. All rights reserved.
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

According to the Refs. [12] and [13], error compensation is divided

into hardware compensation and software compensation. It is
well-known that hardware compensation has economical limitations as
machining costs rise exponentially with the level of machining accuracy
involved [13], whereas software compensation is easier to carry out at
lower costs since only a computer software is needed to compensate er-
rors by correcting numerical control(NC) instructions before executing
compensation instructions [12]. Therefore, many literatures have re-
ported that most researchers used software compensation to improve
machining accuracy. Khan et al. [14] presented a recursive method to
eliminate machine errors from the actual tool path, and then converted
the corrected tool paths into practical compensated NC codes. Zhu et al.
[15] developed a prototype software compensation system to improve
the machining accuracy of a five axis machine tool. The basic compen-
sation idea is to modify the original NC codes to make the actual tool tip
poses reach the desired ones. Cui et al. [12] investigated a software
system to realize error compensation via NC programs reconstructing
according to the predicted errors during virtual machining. Chen et al.
[16] used a software error compensation method to generate NC pro-
grams based on the summation of differential transformation matrix
caused by the geometric errors of each motion axis. Fu et al. [3] described
a strategy to obtain the compensated NC codes through calculating dif-
ferential motion matrix of each axis relative to tool and constructing
Jacobian matrix. Ding et al. [17] proposed a compensation principle to
calculate the corrected NC codes obtained through algebraic operation.
Zhou et al. [18] proposed a new compensation method based on the
topology relation between each axis in kinematic chain of machine tools
to calculate compensated position command of each axis. Table 1 shows Fig. 1. The structure of a NOFAMT.

the contributions of software compensation methods based on the pub-

lished studies in the literature.
As can be observed in the above-mentioned studies, these software
configuration based on the assumption of rigid body motions. Jha et al.
compensation methods only focused on how to calculate the modified
[21] used Denavit-Hartenberg (D-H) method to build a universal error
code of tool path attaching the machine error information without
model to improve the quality of cam profile on a five-axis CNC machine
considering tool cutting trajectory of the machine tool driven by the
tool. Lin et al. [22] introduced a modified Denavit–Hartenberg (MD–H)
modified NC code. However, in practical machining process, the relative
method to model a multi-axis machine tool. Based on the multi-body
motion trajectory of the cutting tool with respect to the workpiece, which
system (MBS), Fan et al. [23] proposed a generalized kinematics error
is known as the form of the machined workpiece, is the most important
modeling method to improve the machining accuracy of NC machine
aspect that people interest in. Therefore, the importance of tool cutting
tools by error compensation. Yang et al. [24] presented a new method to
trajectory should be well concerned. In addition, since the mapping
identify and correct position independent geometric errors (PIGEs)
relationship between the modified code and the tool cutting trajectory
related to five-axis machine tools based on screw theory. Among these
can truly indicate the real response of the machine tool to the modified
methods, MBS can describe the motion relationship among the compo-
NC code, unless it is perfectly represented, the machine tool driven by the
nents of machine tools simply and conveniently [25,26]. Therefore, in
given NC code cannot achieve the desired machining accuracy. In this
this research, HTM and MBS theory were utilized to establish TRMCE.
scenario, authors of this paper have attempted to develop a new iterative
The rest of this paper is arranged as follows. TRMCEs of the tool
methodology to solve this tricky problem.
center position and orientation related to one type non-orthogonal five-
In this paper, the relative motion constraint equation (TRMCE) is the
axis machine tool(NOFAMT) considering geometric error are established
theoretical basis and the crux of software compensation. To establish
based on HTM and MBS theory in Section 2. In order to develop geo-
TRMCE, a systematic and suitable modeling method appears to be very
metric error compensation, a new iterative compensation methodology
necessary [19]. Over the past decades, many published articles have
based on Section 2 is presented in Section 3. In Section 4, experimental
researched the modeling methods and have made noteworthy achieve-
study is carried out to demonstrate the effectiveness of the proposed
ments. Eman et al. [20] utilized homogeneous transform matrix (HTM) to
method, and finally some conclusions are drawn in Section 5.
develop a generalized error model of a multi-axis machine with arbitrary

2. Establishment of TRMCE considering geometric error based on

Table 1
HTM and MBS theory
The contributions of software compensation methods in machine tools.
2.1. Structure of the multi-axis machine tool
References Method Contributions

[5,12–15,18] Iterative (1) To calculate the additional codes attaching the

In this study, a NOFAMT with a nutating head is chosen as the
method machine error information;
(2) To modify the ideal NC codes of tool path using
research object, as shown in Fig. 1. The machine tool consists of machine
the result of step (1). bed (body 1), Y-axis slide carriage (body 2), C-axis rotary table (body 3),
[3,16] Differential (1) To generate NC codes considering the geometric workpiece (body 4), X-axis slide carriage (body 5), Z-axis slide carriage
method errors of each motion axis; (body 6), B-axis nutating head (body 7), spindle (body 8) and cutting tool
(2) To modify the ideal NC codes of tool path based
(body 9). There is an angle between the nutating rotary B-axis and the
on the result of step (1).
[17] Algebraic To directly calculate the corrected NC code of tool axis of cutting tool, which is expressed as α. Note that our method can be
method path considering the machine error information.

81
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

2.2. Topological structure and lower body array Table 2

Low body array for the NOFAMT.

Based on MBS theory, a machine tool can be assumed to be composed Classic body k 1 2 3 4 5 6 7 8 9
of various rigid bodies. Fig. 2 shows the topology structure map of the L0 ðkÞ 1 2 3 4 5 6 7 8 9
NOFAMT, where “Bi” means the i-th body. The lower body array for the L1 ðkÞ 0 1 2 3 1 5 6 7 8
researched machine tool can be figured out as shown in Table 2, and L2 ðkÞ 0 0 1 2 0 1 5 6 7
Table 3 depicts the degrees of freedom (DOFs) between the two adjacent L3 ðkÞ 0 0 0 1 0 0 1 5 6
parts, where “0” means no DOF and “1” means one DOF. As shown in L4 ðkÞ 0 0 0 0 0 0 0 1 5
L5 ðkÞ 0 0 0 0 0 0 0 0 1
Fig. 2, the kinematic chain of the NOFAMT can be divided into two
L6 ðkÞ 0 0 0 0 0 0 0 0 0
branches by considering the body of machine bed as a reference. One is
workpiece branch from machine bed to workpiece, and the other is tool
branch from machine bed to cutting tool. The two adjacent bodies are
Table 3
linked with each other in each branch. The transformation matrix of the
Adjacent body freedom of the NOFAMT.
relative movement between the two adjacent bodies is described as,
Adjacent body X Y Z A B C
½SIJ ¼ ½SIJp ½SIJpe ½SIJs ½SIJse (1) 1–2 0 1 0 0 0 0
2–3 0 0 0 0 0 1
3–4 0 0 0 0 0 0
½SIJðRÞ ¼ ½SIJp ðRÞ½SIJpe ðRÞ½SIJs ðRÞ½SIJse ðRÞ (2) 1–5 1 0 0 0 0 0
5–6 0 0 1 0 0 0
where I and J denote the adjacent higher and lower body respectively, 6–7 0 0 0 0 1 0
and ½SIJp ð½SIJp ðRÞÞ, ½SIJpe ð½SIJpe ðRÞÞ, ½SIJs ð½SIJs ðRÞÞ , ½SIJse ð½SIJse ðRÞÞ 7–8 0 0 0 0 0 0
8–9 0 0 0 0 0 0
describe the relative position (angular position) transformation matrix,
the relative position error(angular position error) transformation matrix,
the relative motion(angular motion) transformation matrix, the relative coordinate system (MCS)fMg , where fxg; fyg; fzgdenote the actual co-
motion error (angular motion error) transformation matrix between the ordinate system of the X; Y; Zdirection respectively, and
rigid body I and the adjacent lower body J respectively. εxy ; εyz ; εxz denote the perpendicularity error of between Xand Y,Yand Z,
X and Z respectively. Taking C rotary axis as an example, its schematic
2.3. Error components definition and setting of the coordinate systems diagram of PIGEs in the C-axis reference coordinate fCg is shown in
Fig. 4, where εxC ; εyC represent the perpendicularity error of between X; Y
According to the Ref. [27], geometric errors are usually composed of axis and C rotary axis respectively.
position-independent geometric errors (PIGEs), and position-dependent Based on the theory of the rigid body motion, each moving part of a
geometric errors (PDGEs). PIGEs don't vary with the moving position machine tool has six DOFs in the Cartesian coordinate system [28].
of the machine tool, but PDGEs are the opposite ones. Fig. 3 is the Hence, each moving axis of the NOFAMT has six PDGEs. Fig. 5 (a) shows
schematic diagram of PIGEs between linear axes in the machine six PDGEs of X-axis, and (b) represents six PDGEs of C-axis, where

Fig. 2. The topology structure map of the NOFAMT.

82
C. Wu et al.
Fig. 3. The schematic diagram of PIGEs between linear axes.
δi ðxÞand εi ðxÞ (i ¼ x; y; z) represent the linear errors and the angular er-
rors of X-axis along the i direction respectively, while δi ðCÞand εi ðCÞ
(i ¼ x; y; z) represent the linear errors and the angular errors of C-axis
along the i direction respectively. Therefore, all the geometric errors of
the NOFAMT are listed, as shown in the Table 4.
For the convenience of this study, the setting of coordinate systems is
conducted as follows: (1) the assumption is made that the body reference
coordinate system (BFCS) of each axis coincides with its motion reference
coordinate system (MFCS) if the machine tool is at the home position (i.e.
the absolute zero point). (2) The direction of each coordinate system
should be identified. Firstly, the machine coordinate system (MCS), i.e.,
the coordinate system of body 1 is selected as the base coordinate system
(BCS), and it is presumed that the MFCS of body 2 and the BCS are in the
same direction. Secondly, let the BCS rotate around axisX and axis Y at
the perpendicularity of εyC and εxC respectively and let the BCS rotate
around axisZ at the perpendicularity of εxy , we can then determine the
MFCS direction of body 3 and 5 respectively. Note that the coordinate
system of body 4 is in line with the coordinate system of body 3. In
addition, the MFCS direction of body 6 can be also confirmed after the
MFCS of body 5 rotates around axisX and axisY at the perpendicularity of
εyz and εxz respectively. Lastly, after the coordinate system, which rotates
around axis X in the MFCS of body 6 at the angle of α rotates around axisX
and axisZ at the perpendicularity of εzB andεxB respectively, the MFCS
direction of body 7 is confirmed. Hence, the MFCS direction of body 8 and
9 can be also identified after the MFCS of body 7 rotates around axis X at
the angle ofα: (3) the position of each coordinate system should be
confirmed. Firstly, let the machine tool go back to the home position, then
the geometrical center (oC ) of the end surface in body 3 is selected as the
origin of the BFCS and MFCS in body 1,2,3,5 and 6, while the origin of the
BFCS and MFCS in body 7, 8 and 9 is set at the intersection point (oB ) of
axis center line in body 6 and 9, as shown in Fig. 1.
2.4. Establishment of TRMCE
In MBS theory, the transformation relationship between the two
" #
 T Y
t¼1
 
PT ¼ PxT PyT PzT 1 ¼ SLt ðjÞLt1 ðjÞ p SLt ðjÞLt1 ðjÞ pe SLt ðjÞLt1 ðjÞ s SLt ðjÞLt1 ðjÞ se rt
t¼n;Ln ð6Þ¼1
¼ ½S15p ½S15pe ½S15s ½S15se ½S56p ½S56pe ½S56s ½S56se ½S67p ½S67pe ½S67s
½S78p ½S78pe ½S78s ½S78se ½S89p ½S89pe ½S89s ½S89se rt
International Journal of Machine Tools and Manufacture 124 (2018) 80–98
Fig. 4. The schematic diagram of PIGEs between X,Y axis and C rotary axis.
adjacent bodies can be represented using a 44 HTM.
Supposing that the coordinate of the cutting tool tip position and
orientation in the tool coordinate system (TCS) can be expressed as rt ¼
ð 0 0 L 1 ÞT and vt ¼ ð 0 0 1 0 ÞT respectively, meanwhile,
the coordinate of the cutting tool tip position and orientation in the
workpiece coordinate system (WCS) can be expressed as


rw ¼ xw yw zw 1 T and vw ¼ ð xv yv zv 0 ÞT respectively.
Moreover, at the home position, the coordinate of the workpiece cutting
point in C-axis BFCS and the coordinate of oB in the MCS can be written as
 T  T
qwx qwy qwz 1 and q7x q7y q7z 1 respectively. Where L
indicates the length of the cutting tool between oB and the center point of
tool tip.
As discussed in Section 2.3, the workpiece is fixed on the C axis. And
the spindle and cutting tool are fixed on the B axis. So the motion or
motion error transformation matrix between the workpiece and C axis
can be considered as an identity matrix. Similarly, the motion or motion
error transformation matrix between the B axis, spindle and cutting tool
can also be considered as an identity matrix respectively. In addition, as
shown in Fig. 6, which is the schematic diagram of the BFCS relation
between body 6, 7 and 8, there is an included angle α between the B-axis
and the spindle. After rotate the BFCS of Z axis around X direction at the
angle of α, the BFCS of B axis can be identified. While the rotation center
of B axis is Z direction. Hence, the position, position error, motion and
motion error transformation matrix between Z axis and B axis can be
obtained respectively. Similarly, the BFCS of spindle can be obtained
through rotating the BFCS of B axis around X direction at the angle of
α:Then, the position transformation matrix between the B axis and
spindle can be given. Followed by analogy, the rest transformation
matrices between the adjacent bodies can be determined. Tables 5–8
show the transformation matrices for all of the adjacent bodies.
In tool branch, the cutting tool tip position can be described to MCS,
as shown in Eq. 3.
    
(3)
83
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Fig. 5. The schematic diagram of PDGEs: (a) X-axis, (b) C-axis.

Table 4
Geometric errors of the NOFAMT.

Linear error Angular error

Axis x direction y direction z direction Around x Around y Around z

PDGEs X δx ðxÞ δy ðxÞ δz ðxÞ εx ðxÞ εy ðxÞ εz ðxÞ

Y δx ðyÞ δy ðyÞ δz ðyÞ εx ðyÞ εy ðyÞ εz ðyÞ
Z δx ðzÞ δy ðzÞ δz ðzÞ εx ðzÞ εy ðzÞ εz ðzÞ
B δx ðBÞ δy ðBÞ δz ðBÞ εx ðBÞ εy ðBÞ εz ðBÞ
C δx ðCÞ δy ðCÞ δz ðCÞ εx ðCÞ εy ðCÞ εz ðCÞ
PIGEs εyz ; εBz ; εyC εxz ; εxC εxy ; εxB

Fig. 6. The schematic diagram of the BFCS relation between body 6, 7 and 8.

where PiT ði ¼ x; y; zÞ denotes the component of PT in i direction. high order terms are much smaller than other terms, the high order terms
Then, Eqs.( (4)–(6) can be obtained by substituting the corresponding are neglected in this process [29].
transformation matrices shown in Tables 5–8 into Eq. (3). Because the

84
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Table 5
Position and position error transformation matrix of the NOFAMT.

Adjacent body Position transformation matrix Position error transformation matrix

1–2 ½S12p ¼ I44 ½S12pe ¼ I44

Y-axis
2 3
2–3 ½S23p ¼ I44 1 0 εxC 0
C-axis 6 0 1 εyC 07
½S23pe ¼64 εxC
7
εyC 1 05
0 0 0 1
2 3
3–4 1 0 0 qwx ½S34pe ¼ I44
workpiece 60 1 0 qwy 7
½S34p ¼ 6
40
7
0 1 qwz 5
0 0 0 1
2 3
1–5 ½S15p ¼ I44 1 εxy 0 0
X-axis 6 εxy 1 0 07
½S15pe ¼64 0
7
0 1 05
0 0 0 1
2 3
5–6 ½S56p ¼ I44 1 0 εxz 0
Z-axis 6 0 1 εyz 0 7
½S56pe ¼64 εxz εyz
7
1 05
0 0 0 1
2 3 2 32 3
6–7 1 0 0 q7x 1 εxB 0 0 1 0 0 0
60 q7y 7 6 εxB εBz 0 7 6 07
B-axis
½S67p ¼ 6
1 0 7 ½S67pe ¼6
1 76 0 cos α sin α 7
40 0 1 q7z 5 4 0 εBz 1 0 5 4 0 sin α cos α 05
0 0 0 1 0 0 0 1 0 0 0 1
2 3
7–8 1 0 0 0 ½S78pe ¼ I44
spindle 60 cos α sin α 07
½S78p ¼ 6
40
7
sin α cos α 05
0 0 0 1
8–9 ½S89p ¼ I44 ½S89pe ¼ I44
tool

Table 6
Motion and motion error transformation matrix of the NOFAMT.

Adjacent body Motion transformation matrix Motion error transformation matrix

2 3 2 3
1–2 1 0 0 0 1 εz ðyÞ εy ðyÞ δx ðyÞ
Y-axis 60 1 0 y 7 6 εz ðyÞ 1 εx ðyÞ δy ðyÞ 7
½S12s ¼ 6
40 0 1 05
7 ½S12se ¼ 6
4 εy ðyÞ εx ðyÞ
7
1 δz ðyÞ 5
0 0 0 0 0 0 0 1
2 3 2 3
2–3 cos C sin C 0 0 1 εz ðCÞ εy ðCÞ δx ðCÞ
C-axis 6 sin C cos C 0 07 6 εz ðCÞ 1 εx ðCÞ δy ðCÞ 7
½S23s ¼ 6
4 0
7 ½S23se ¼ 6 7
0 1 05 4 εy ðCÞ εx ðCÞ 1 δz ðCÞ 5
0 0 0 1 0 0 0 1
3–4 ½S34s ¼ I44 ½S34se ¼ I44
workpiece
2 3 2 3
1–5 1 0 0 x 1 εz ðxÞ εy ðxÞ δx ðxÞ
X-axis 60 1 0 07 6 εz ðxÞ 1 εx ðxÞ δy ðxÞ 7
½S15s ¼640 0
7 ½S15se ¼6 7
1 05 4 εy ðxÞ εx ðxÞ 1 δz ðxÞ 5
0 0 0 0 0 0 0 1
2 3 2 3
5–6 1 0 0 0 1 εz ðzÞ εy ðzÞ δx ðzÞ
Z-axis 60 1 0 07 6 εz ðzÞ 1 εx ðzÞ δy ðzÞ 7
½S56s ¼640 0
7 ½S56se ¼6 7
1 z5 4 εy ðzÞ εx ðzÞ 1 δz ðzÞ 5
0 0 0 0 0 0 0 1
2 3 2 3
6–7 cos B sin B 0 0 1 εz ðBÞ εy ðBÞ δx ðBÞ
B-axis 6 sin B cos B 0 07 6 εz ðBÞ 1 εx ðBÞ δy ðBÞ 7
½S67s ¼64 0
7 ½S67se ¼6 7
0 1 05 4 εy ðBÞ εx ðBÞ 1 δz ðBÞ 5
0 0 0 1 0 0 0 1
7–8 ½S78s ¼ I44 ½S78se ¼ I44
spindle
8–9 ½S89s ¼ I44 ½S89se ¼ I44
tool

   
PxT ¼ δx ðxÞ þ δx ðzÞ þ q7x þ x þ z εy ðxÞ þ εxz þ δx ðBÞcosB þ δy ðBÞsin B þ q7z εy ðxÞ þ εy ðzÞ þ εxz
q7y ðεz ðxÞ þ εxy Þ þ Lðsin α sin B  ðcos2 α þ sin2 αcosBÞðεy ðxÞ þ εy ðzÞ þ εxz Þ  cosαðcosBεy ðBÞ
þsin Bεx ðBÞÞ  cosα sin αðεz ðxÞ þ εz ðzÞ þ εxB þ εxy Þ þ sin αcosBεz ðBÞ þ cosα sin αcosBðεz ðxÞ þ εz ðzÞ þ εxB þ εxy ÞÞ (4)

85
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Table 7
Angular position and angular position error transformation matrix of the NOFAMT.

Adjacent body Angular position transformation matrix Angular position error transformation matrix

1–2 ½S12p ðRÞ ¼ I44 ½S12pe ðRÞ ¼ I44

Y-axis
2 3
2–3 ½S23p ðRÞ ¼ I44 1 0 εxC 0
C-axis 6 0 1 εyC 07
½S23pe ðRÞ ¼ 6
4 εxC
7
εyC 1 05
0 0 0 1
3–4 ½S34p ðRÞ ¼ I44 ½S34pe ðRÞ ¼ I44
workpiece
2 3
1–5 ½S15p ðRÞ ¼ I44 1 εxy 0 0
X-axis 6 εxy 1 0 07
½S15pe ðRÞ ¼ 6
4 0
7
0 1 05
0 0 0 1
2 3
5–6 ½S56p ðRÞ ¼ I44 1 0 εxz 0
Z-axis 6 0 1 εyz 0 7
6
½S56pe ðRÞ ¼ 4 7
εxz εyz 1 05
0 0 0 1
2 32 3
6–7 ½S67p ðRÞ ¼ I44 1 εxB 0 0 1 0 0 0
6 εxB εBz 0 7 6 07
B-axis
½S67pe ðRÞ ¼ 6
1 76 0 cos α sin α 7
4 0 εBz 1 0 54 0 sin α cos α 05
0 0 0 1 0 0 0 1
7–8 ½S78p ðRÞ ¼ I44 ½S78pe ðRÞ ¼ I44
spindle
8–9 ½S89p ðRÞ ¼ I44 ½S89pe ðRÞ ¼ I44
tool

Table 8
Angular motion and angular motion error transformation matrix of the NOFAMT.

Adjacent body Angular motion transformation matrix Angular motion error transformation matrix
2 3
1–2 ½S12s ðRÞ ¼ I44 1 εz ðyÞ εy ðyÞ 0
Y-axis 6 εz ðyÞ 1 ε ðyÞ 0 7
½S12se ðRÞ ¼ 6
4 εy ðyÞ εx ðyÞ
x 7
1 05
0 0 0 1
2 3 2 3
2–3 cos C sin C 0 0 1 εz ðCÞ εy ðCÞ 0
C-axis 6 sin C cos C 0 07 6 εz ðCÞ 1 ε ðCÞ 0 7
½S23s ðRÞ ¼ 6
4 0
7 ½S23se ðRÞ ¼ 6 x 7
0 1 05 4 εy ðCÞ εx ðCÞ 1 05
0 0 0 1 0 0 0 1
3–4 ½S34s ðRÞ ¼ I44 ½S34se ðRÞ ¼ I44
workpiece
2 3
1–5 ½S15s ðRÞ ¼ I44 1 εz ðxÞ εy ðxÞ 0
X-axis 6 εz ðxÞ 1 εx ðxÞ 07
½S15se ðRÞ ¼ 6
4 εy ðxÞ εx ðxÞ
7
1 05
0 0 0 1
2 3
5–6 ½S56s ðRÞ ¼ I44 1 εz ðzÞ εy ðzÞ 0
Z-axis 6 εz ðzÞ 1 εx ðzÞ 07
½S56se ðRÞ ¼ 6
4 εy ðzÞ εx ðzÞ
7
1 05
0 0 0 1
2 3 2 3
6–7 cos B sin B 0 0 1 εz ðBÞ εy ðBÞ 0
B-axis 6 sin B cos B 0 07 6 εz ðBÞ 1 εx ðBÞ 07
½S67s ðRÞ ¼ 6
4 0
7 ½S67se ðRÞ ¼ 6 7
0 1 05 4 εy ðBÞ εx ðBÞ 1 05
0 0 0 1 0 0 0 1
2 3
7–8 1 0 0 0 ½S78se ðRÞ ¼ I44
spindle 6 0 cos α sin α 07
½S78s ðRÞ ¼ 6
4 0 sin α cos α
7
05
0 0 0 1
8–9 ½S89s ðRÞ ¼ I44 ½S89se ðRÞ ¼ I44
tool

     
PyT ¼ δy ðxÞ þ δy ðzÞ þ q7y þ xεxy  z εx ðxÞ þ εyz  q7z εx ðxÞ þ εx ðzÞ þ εyz þ q7x εz ðxÞ þ εz ðzÞ þ εxy
sin αδz ðBÞ þ cosαðcosBδy ðBÞ þ sin Bδx ðBÞÞ þ Lðsin2 αεx ðBÞ þ cosα sin α þ cos2 αðεx ðxÞ þ εx ðzÞ
þεBz þ εyz þ cosBεx ðBÞ  sin Bεy ðBÞÞ þ sin2 αcosBðεx ðxÞ þ εx ðzÞ þ εyz þ εBz Þ þ sin α sin Bðεz ðxÞ
þεz ð0zÞ þ εxy þ εxB ÞÞ (5)

86
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

 
PzT ¼ δz ðxÞ þ δz ðzÞ þ q7z þ z þ cosαδz ðBÞ þ q7y εx ðxÞ þ εx ðzÞ þ εyz
     where PiW and ViW ði ¼ x; y; zÞ denote the component of PW and VW in i
 q7x εy ðxÞ þ εy ðzÞ þ εxz þ sin α cosBδy ðBÞ þ sin Bδx ðBÞ þ L
  direction respectively.
 cos2 α  sin2 αcosB þ cosα sin α εBz  εx ðBÞ þ εyz Substituting the corresponding transformation matrices shown in
  
 sin α sin B εy ðxÞ þ εy ðzÞ þ εxz þ cosα sin αcosB εxB  εx ðxÞ Tables 5–8 into Eqs.( (11) and (12), the following equations can be ob-
  tained, as shown in Eqs.(13)–(18).
 εx ðzÞ  εBz  εyz  cosα sin α sin Bεy ðBÞ
(6)
Similarly, the cutting tool tip orientation can also be described to
MCS, as shown in Eq. (7).

VT "¼ ð VxT VyT VzT 0 ÞT #

Y
t¼1
 t   t   t   t 
¼ ð SL ðjÞL ðjÞ p ðRÞ SL ðjÞL ðjÞ pe ðRÞ SL ðjÞL ðjÞ s ðRÞ SL ðjÞL ðjÞ se ðRÞÞ vt
t1 t1 t1 t1

t¼n;Ln ð6Þ¼1 (7)

¼ ½S15p ðRÞ½S15pe ðRÞ½S15s ðRÞ½S15se ðRÞ½S56p ðRÞ½S56pe ðRÞ½S56s ðRÞ½S56se ðRÞ
½S67p ðRÞ½S67pe ðRÞ½S67s ðRÞ½S67se ðRÞ½S78p ðRÞ½S78pe ðRÞ½S78s ðRÞ½S78se ðRÞ
½S89p ðRÞ½S89pe ðRÞ½S89s ðRÞ½S89se ðRÞvt

PxW ¼ δx ððyÞÞ  ðqwy þ yw Þðsin C þ cos Cεz ðCÞ þ cos Cεz ðyÞÞ þ ðqwx
where ViT ði ¼ x; y; zÞ denotes the component of VT in i direction.
Then, Eqs.( (8)–(10) can be obtained based on Eq. (7) as follows. þ xw Þðcos C  sin Cεz ðCÞ  sin Cεz ðyÞÞ þ ðqwz þ zw Þðεx ðCÞ
þ εy ðyÞ þ cos Cεy ðCÞ þ sin Cεx ðCÞÞ þ cos Cδx ðCÞ  sin Cδy ðCÞ
VxT ¼ sin Bðsinα  cos αεx ðBÞÞ þ cos Bðsinαεz ðBÞ  cos αεy ðBÞ
  (13)
 sin2 αðεy ðxÞ þ εy ðzÞ þ εxz Þ þ cos αsinα εz ðxÞ þ εz ðzÞ þ εxB þ εxy
 
 cos2 α εy ðxÞ þ εy ðzÞ þ εxz  cos αsinαðεz ðxÞ þ εz ðzÞ þ εxB þ εxy Þ PyW ¼ δx ðyÞ þ y þ ðqwx þ xw Þðsin C þ cos Cεz ðCÞ þ cos Cεz ðyÞÞ þ ðqwy
(8) þ yw Þðcos C  sin Cεz ðCÞ  sin Cεz ðyÞÞ  ðqwz þ zw Þðεx ðyÞ
þ εy ðCÞ þ cos Cδx ðCÞ  sin Cδy ðCÞÞ þ cos Cδy ðCÞ þ sin Cδx ðCÞ
VyT ¼ cos2 αðεx ðxÞ þ εx ðzÞ þ εBz þ εyz Þ þ sin2 αεx ðBÞ þ cos αsinα
 (14)
þ cos Bðcos αsinα þ cos2 αεx ðBÞ þ sin2 α εx ðxÞ þ εx ðzÞ þ εBz
  
þ εyz þ sin B sinαðεz ðxÞ þ εz ðzÞ þ εxB þ εxy  cos2 αεy ðBÞ PzW ¼ δz ðCÞ þ δz ðyÞ þ qwz þ zw  ðqwx þ xw Þðεy ðCÞ þ cos Cεx ðCÞ
þ cos αsinαεz ðBÞÞ þ cos Cεy ðyÞ þ sin Cεx ðyÞ þ sin Cεy ðCÞÞ þ ðqwy þ yw Þðεx ðCÞ
(9) þ cos Cεx ðyÞ þ cos Cεy ðCÞ þ sin Cεx ðCÞ þ sin Cεy ðyÞÞ
(15)
VzT ¼ cos αsinαðεx ðxÞ  εx ðBÞ þ εx ðzÞ þ εBz þ εyz Þ  cos2 α
  
 cos Bðsin2 α þ cos αsinα εx ðxÞ  εx ðBÞ þ εx ðzÞ þ εBz þ εyz VxW ¼ xv ðcosC  sin Cεz ðCÞ  sin Cεz ðyÞÞ þ zv εx ðCÞ þ εy ðyÞ

 sin Bsinαðεy ðxÞ þ εy ðzÞ þ εxz  sinαεz ðBÞ þ cos αεy ðBÞÞ þ cosCεy ðCÞ þ sin Cεx ðCÞ  yv ðsin C þ cosCεz ðCÞ
(10) þ cosCεz ðyÞÞ (16)
In addition, in workpiece branch, the required tool position and
orientation of the workpiece cutting point in the MCS can be expressed in
Eqs.( (11) and (12) respectively.

" #
Y
t¼1
 t  t   t  t 
PW ¼ ð PxW PyW PzW 1Þ ¼
T
ð SL ðjÞL ðjÞ p SL ðjÞL ðjÞ pe SL ðjÞL ðjÞ s SL ðjÞL ðjÞ se Þ rw
t1 t1 t1 t1

t¼n;Ln ð4Þ¼1
(11)
¼ ½S12p ½S12pe ½S12s ½S12se ½S23p ½S23pe ½S23s ½S23se ½S34p ½S34pe ½S34s ½S34se rw

 T
VW ¼" VxW VyW VzW 0 #
Y
t¼1
   t   t   t  
¼ SL ðjÞL ðjÞ p ðRÞ SL ðjÞL ðjÞ pe ðRÞ SL ðjÞL ðjÞ s ðRÞ SL ðjÞL ðjÞ se ðRÞ vw
t t1 t1 t1 t1
(12)
t¼n;Ln ð4Þ¼1
¼ ½S12p ðRÞ½S12pe ðRÞ½S12s ðRÞ½S12se ðRÞ½S23p ðRÞ½S23pe ðRÞ½S23s ðRÞ½S23se ðRÞ
½S34p ðRÞ½S34pe ðRÞ½S34s ðRÞ½S34se ðRÞÞ vw

87
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98


VyW ¼ yv ðcosC  sin Cεz ðCÞ  sin Cεz ðyÞÞ  zv εx ðyÞ þ εy ðCÞ Evi ði ¼ x; y; zÞare displayed in Eqs. A1 and A2 in Appendix A respectively.
 Epi denotes the component of TRMCE of the tool tip position in i direc-
þ cosCεx ðCÞ  sin Cεy ðCÞ þ xv ðsin C þ cosCεz ðCÞ
tion. Evi denotes the component of TRMCE of tool orientation vector in
þ cosCεz ðyÞÞ (17) i direction.

VzW ¼ zv  xv ðεy ðCÞ þ cosCεx ðCÞ þ cosCεy ðyÞ  sin Cεx ðyÞ 3. Iterative compensation methodology based on TRMCE

 sin Cεy ðCÞÞ þ yv ðεx ðCÞ þ εx ðyÞ þ cosCεy ðCÞ þ sin Cεx ðCÞ
In this work, the core idea of the new iterative compensation meth-
þ sin Cεy ðyÞÞ odology is that we only focus on how to obtain the actual NC codes
(18) controlling the machine tool to achieve the desired machining accuracy
without concerning the compensated values in machining process. In
During the practical machining process, to obtain the desired accu- order to implement the methodology, the mutual mapping relationships
racy of workpiece, the position and orientation of the cutting tool tip in between tool path, NC command and tool cutting trajectory based on
tool branch should be coincident with the required tool position and TRMCE should be established respectively.
orientation of the workpiece cutting point in workpiece branch at any Tool path refers to the theoretical tool center position curve, which is
moment in MCS, i.e.,PT ¼ PW ; VT ¼ VW . Therefore, TRMCE of the tool tip planned in WCS according to the design drawings, and has nothing to do
position and tool orientation vector related to the researched NOFAMT with the motion of a machine tool. Whereas, tool cutting trajectory refers
can be gotten with Eqs.( (19) and (20) respectively. to motion trajectory of the cutting tool in WCS, which can be divided into
2 3 ideal tool cutting trajectory and actual tool cutting trajectory. Ideal tool
Epx
6 Epy 7 cutting trajectory is the tool cutting trajectory obtained by theoretical
Ep ¼ 6 7
4 Epz 5 ¼ 0 (19) calculation based on the nominal motion model of a machine tool, while
1 actual tool cutting trajectory is the tool cutting trajectory generated in the
practical machining process of a machine tool. Fig. 7 shows the schematic
2 3 diagram of the relationships between tool path, NC command and tool
Evx cutting trajectory.
6 Evy 7
Ev ¼ 6 7
4 Evz 5 ¼ 0 (20)
3.1. Mapping relationship between tool path and NC command
0
Ignoring geometric errors in Eqs.( (19) and (20), the mapping rela-
where Ep and Ev represent the TRMCE of the tool tip position and tool
tionship between tool path and the ideal NC command can be repre-
orientation vector respectively. The algebraic expressions of Epi and
sented as,

The design drawings of parts

Tool path

Mapping model of the actual machine Mapping model of the idea machine
tool considering geometric error tool not considering geometric error

The modified NC command The idea NC command

The actual machine The idea machine

The actual machine tool
tool tool

Tool cutting trajectory Tool cutting trajectory Tool cutting trajectory

without error with error without error

Fig. 7. The schematic diagram of the mapping relationships between tool path, NC command and tool cutting trajectory.

88
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

2 3
2 3 qwx cos C  qwy sin C  L sin α sin B  q7x machining. Hence, the mapping relationship between tool path and the
x 6 7
6 y 7 6 qwx sin C  qwy cos C  L cos α sin α cos B 7 actual NC command should be established.
6 7 ¼ 6 þL cos α sin α þ q7y 7 Taking into account the geometric errors and simplifying Eqs.( (19)
4z 5 6 7
4 qwz þ Lsin2 α cos B þ L cos α  q7z 5 and (20), the mapping relationship between tool path in WCS and the
1
0 actual NC command of the tool tip position in MCS is represented by Eq.
2 32 3 (23), and the mapping relationship between tool orientation vector in
cos C sin C 0 0 xw
6 sin C cos C 0 0 76 yw 7 WCS and the actual NC command of the tool Angular position in MCS is
þ6 4 0
76 7 (21)
0 1 0 54 zw 5 represented by Eqs.( (24) and (25).
0 0 0 1 1

2q 7x ðεz ðxÞ
þ εz ðzÞÞ þ q7y  q7z εx ðzÞ þ δy ðxÞ  δy ðyÞ þ δy ðzÞ  cos Cδy ðCÞ  sin α 3
δz ðBÞ  sin Cδx ðCÞ þ cosαcosBδx ðBÞ þ cosα sin Bδx ðBÞ þ Lðcosα sin α þ cos2 α
6 7
ðεx ðzÞ þ εBz Þ þ sin2 αεx ðBÞ þ cos2 αcosBεx ðBÞ þ sin2 αcosBðεx ðzÞ þ εBz Þ  cos2 α
6 sin Bεy ðBÞ þ cosα sin αcosB þ sin α sin Bðεz ðxÞ þ εz ðzÞÞÞ
7
2 3 6 q þ q ðε ðxÞ þ ε ðzÞ þ ε Þ  q ε ðzÞ þ δ ðyÞ  δ ðxÞ  δ ðzÞ  cosBδ ðBÞ 7
x 6 7x 7y z z xy 7z y
7
6 y 7 6 þcos Cδ ðCÞ þ sin Bδ ðBÞ þ sin Cδ ðCÞ þ Lðsin α sin B þ cos αε ðzÞ þ sin α 7
x x y
x x x
2
y
x
2

6 7 ¼ cosBε ðzÞ þ cosαcosBε ðBÞ þ cosα sin Bε ðBÞ þ cosα sin αðε ðxÞ þ ε ðzÞþ
4z5 6 y y x
7 z z

1 6 ε þ ε Þ  sin αcosBε ðBÞ  cosα sin αcosBðε ðxÞ þ ε ðzÞ þ ε þ ε ÞÞ

xB xy z
7z z xB xy

6 q ðε ðxÞ þ ε ðzÞ þ ε Þ  q  q ðε ðxÞ þ ε ðzÞ þ ε Þ þ δ ðCÞ þ δ ðyÞ 7

4 δ ðxÞ  δ ðzÞ  cosαδ ðBÞ  sin αðcosBδ ðBÞ þ sin Bδ ðBÞÞ  Lðcos α 5
7x
z
y
z
y xz
z
7z 7y x
y
x yz
x
z z
2
2
sin αcosB þ cosα sin αðεBz  εx ðBÞ þ εyz Þ  sin α sin Bðεy ðxÞ þ εy ðzÞ þ εxz Þ (23)
þcosα sin αcosBðεxB  εx ðxÞ  εx ðzÞ  εBz  εyz Þ  cosα sin α sin Bεy ðBÞÞ
0
2 sin C þ cos Cεxy  cos C  sin Cεxy ðεyz þ εx ðxÞ  εx ðyÞ 3
cos Cεz ðCÞ þsin Cεz ðCÞþ εy ðCÞ  cos Cεx ðCÞ 0
6 7
6 cos Cεz ðyÞ sin Cεz ðyÞ þsin Cεy ðCÞÞ 7
6 72 3
6 ðεxz þ εy ðxÞ  εx ðCÞ 7
6 cos C  sin Cεz ðCÞ ðsin C þ cos C 7 qwx þ xw
6 εy ðyÞ  cos Cεy ðCÞ 0 76
6 sin Cεz ðyÞ εz ðCÞ þ cos Cεz ðyÞÞ 76 qwy þ yw 7
7
þ6 sin Cεx ðCÞÞ 74
6 7 qwz þ zw 5
6 ðεy ðCÞ þ cos C εx ðCÞ þ cos C 7
6 7 1
6 εx ðCÞ þ cos C εx ðyÞ þ cos C 7
6 1 07
4 εy ðyÞ þ sin C εy ðCÞ þ sin C 5
εx ðyÞ þ sin Cεy ðCÞÞ εx ðCÞ þ sin Cεy ðyÞ
0 0 0 1

2 3 2 3
cos C 0 temp1 cos C þ temp2 sin C ¼ temp3 (24)
6 sin C 7 6 0 7
6 7 6 7
4 cos B 5 ¼ 4 cot2 α 5 temp4 cos B ¼ temp5 (25)
0 0
2 3
sin α sin B cos α sin αð1  cos BÞ 0 0 where, the algebraic expressions of tempi(i ¼ 1, 2, 3, 4, 5) are displayed in
6 cos α sin αð1  cos BÞ sin α sin B 0 07 Eq. B1 in Appendix B.
þ4 6 7
0 0 csc2 α 05 Then, Eq. (26) is given by the simplification of Eq. (24).
0 0 0 1
2 xv 3 temp1 temp2
cos C pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ sin C pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 temp1 þ temp2
2 2 temp12 þ temp22
6 xv þ yv 7
6 7
6 yv 7 temp3
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
6 2 7 (26)
 6 x þ y2 7 temp12 þ temp22
6 v v7
6 7
4 zv 5 Assuming C axis can rotate 3600 (0  C  3600 ), and B axis can rotate
0 1800 (0  B  1800 ), then the following conditions can be generated.
(22)
(1) When temp2  0, then θ ¼ arccos pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
temp1
; when temp2 < 0,
2 2 temp1 þtemp2
where x; y; z; B; C indicate the idea NC command in MCS, xw ; yw ; zw
indicate the position coordinates of tool path in WCS, xv ; yv ; zv indicate then θ ¼ 3600  arccos pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
temp1
.
2 temp1 þtemp2
2

the tool orientation vector in WCS. (2) When 3600  C  θ  1800 ,

However, the ideal NC command without considering geometric er- thenC ¼ 360 þ arccos
0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
temp3
þ θ; when
rors of the machine tool can't meet the demands of the practical
2 temp1 þtemp2
2

89
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

1800  C  θ  0, then C ¼ arccos pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

temp3
þ θ ; when
2 temp1 þtemp2
2

0  C  θ  1800 , then C ¼ arccos pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

temp3
 θ; when
temp1 þtemp2
2 2

1800  C  θ  3600 , thenC ¼ 3600  arccos pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

temp3
 θ.
temp12 þtemp22
temp5
(3) B ¼ arccos temp4 .

3.2. Mapping relationship between NC command and tool cutting

trajectory

In the practical machining process, actual tool cutting trajectory

should be concerned, which will seriously affect the machining accuracy
of the machine tool. Therefore, the mapping relationship between NC
command and tool cutting trajectory should be established. Based on Eq.
(19), Eq. (27) can be obtained through substituting xw' ; yw' ; zw' for
xw ; yw ; zw respectively. Then, mapping relationship between the NC
command and the actual tool tip cutting trajectory is generated by Fig. 8. The relative position relationship between cutting tool and workpiece in WCS.
simplifying Eq. (27), as shown in Eq. (28).

8
<x 0
w ¼ ux þ ðcos C  sin Cðεz ðCÞ þ εz ðyÞ  εxy ÞÞx  ðsin C þ cos Cðεz ðCÞ þ εz ðyÞÞÞy þ ðsin Cðεx ðyÞ þ εy ðCÞ
εx ðxÞ  εyz Þ  cosCðεx ðCÞ þ εy ðyÞ þ εy ðxÞ þ εxz Þ  εy ðCÞÞz
y0w ¼ uy  ðsin C þ cos Cðεz ðCÞ þ εz ðyÞ  εxy ÞÞx  ðcos C  sin Cðεz ðCÞ þ εz ðyÞÞÞy þ ðcosCðεx ðyÞ þ εy ðCÞ (27)
: εx ðxÞ  εyz Þ þ sin Cðεx ðCÞ þ εy ðyÞ þ εy ðxÞ þ εxz Þ þ εx ðCÞÞz
z0w ¼ uz þ ðεx ðCÞ þ εy ðyÞ þ cos Cεy ðCÞ þ sin Cεx ðCÞÞx þ ðεy ðCÞ þ εx ðyÞ þ cos Cεx ðCÞ  sin Cεy ðCÞÞy þ z

where xw' ; yw' ; zw' indicate the position coordinates of the actual tool tip
2 3 2 2 3
3
x'w ux x cutting trajectory, xv' ; yv' ; zv' indicate the actual tool orientation vector, the
6 y' 7 6 uy 7 6 7 algebraic expressions of ui (i ¼ 1; 2; 3; 4; 5),vj (j ¼ 1; 2; 3; 4; 5; 6) and uk
6 'w 7 ¼ 6 7 þ Tw 6 y 7 (28)
4z 5 4 uz 5 4z5 (k ¼ x; y; z) are displayed in Eqs. C1 and C2 in Appendix C respectively.
w
1 0 1 Tw ,Tv1 ,Tv2 are displayed in Appendix D.
Eqs.(27) and (28), from the physical sense, can provide the driving
Similarly, Eq. (29) is obtained by using Eq. (20) and substituting
model of an actual machine tool. Using the model, the tool motion tra-
xv' ; yv' ; zv' for xv ; yv ; zv respectively. Then, the mapping relationship be- jectory of the machine tool driven by the given NC command can be
tween the NC command of the tool angular position and the actual tool really reproduced.
orientation is represented in Eq. (30). Furthermore, the model can also be used to obtain volumetric error of
any position point in the workpiece directly. Fig. 8 shows the relative

8
>
> x0v ¼ cos2 αεy ðCÞ þ u1 cos B cos C þ ðsinα  cos αεx ðBÞÞsin B cos C þ u2 sin B sin Cþ
>
>
>
> u3 cos B sin C þ v1 cos C þ v2 sin C þ sin2 αεy ðCÞcos
 B 
>
>
< yv ¼ cos αεx ðCÞ  sin2 α þ u4 cos B cos C þ u5 sin B cos C  εx ðCÞ þ εy ðyÞ cos B sin C
0 2

 þv3 cos C þ v4 sin C þ εx ðCÞcos B  (29)

>
>
>
> z0v ¼ cos α sin α εx ðxÞ  εx ðBÞ þ εx ðzÞ þ εBz þ εyz  εx ðyÞ  εy ðCÞ  cos2 α þ cos α sin αεx ðCÞ
>
>
>
> cos B cos C þ sin αεy ðCÞsin B cos C þ sin αεx ðCÞsin B sin C  cos α sin αεy ðCÞcos B sin C
:
cos α sin αεx ðCÞcos C þ cos α sin αεy ðCÞsin C þ v5 cos B þ v6 sin B

position relationship between cutting tool and workpiece in WCS (i.e.

cos2 αεy ðCÞ Ow  Xw Yw Zw ). Where PTW denotes the position coordinate of tool tip
2 3 2 3 2 3
x0v ðcos αεx ðCÞ þ sin2 αÞ
2 cos B cos C point in WCS. VTW denotes the cutting tool orientation coordinate in
6 y0v 7 6 cos α sin αðεx ðxÞ  εx ðBÞ 7 6 7
7 þ Tv1 6 sin B cos C 7
WCS. PPT denotes the position coordinate of the point P in the cutting tool
4 0 5¼6
4 þεx ðzÞ þ εBz þ εyz  εx ðyÞ 5 4 sin B sin C 5 center. D denotes the distance between the tool tip point and the point P
zv
0 εy ðCÞÞ  cos2 α cos B sin C in the direction of the cutting tool center. O denotes the position of the
0 tool tip point. It is obvious that the following equations can be obtained.
2 3
cos C
6 sin C 7 PPT ¼ PTW þ D*VTW (31)
6
þ Tv2 4 7 (30)
cos B 5 When O represents the position of the actual tool tip point, then
sin B  T  T
PTW ¼ xw' yw' zw' , and VT ¼ xv' yv' zv' . While when O

90
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

represents the position of the ideal tool tip point, then

T T
PTW ¼ ð xw yw zw Þ , and VT ¼ ð xv yv zv Þ .
Hence, according to Eq. (31), the coordinates of actual position and
ideal position for the point P in the cutting tool center can be obtained,
 T
which is written as PPTactual ¼ xw' þ Dxv' yw' þ Dyv' zw' þ Dzv'
andPPTideal ¼ ð xw þ Dxv yw þ Dyv zw þ Dzv ÞT respectively. Then, the
volumetric error of the point P can be obtained, and finally expressed
as follow.
8 2 3 2 0 3
>
> Ex xw þ Dx0v  xw  Dxv
>
< Ev ¼ PPTactual  PPTideal ¼ 4 Ey 5 ¼ 4 y0w þ Dy0v  yw  Dyv 5
> Ez z0w þ Dz0v  zw  Dzv (32)
>
> qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
: E ¼ E2 þ E2 þ E2
x y z

whereEv denotes the coordinate vector of the volumetric error. Ex ,Ey ,Ez
denote the component of Ev in x; y; zdirection respectively. E denotes the
volumetric error size.

3.3. The iterative calculation method of the NC command corrected values

In this part, the corrected values of the actual NC command are

calculated by utilizing the aforementioned mapping relationships.
Whereas, it is difficult to calculate the command values directly and
accurately due to the nonlinearity of mapping relationships and the
variability of geometric errors at diverse positions. Therefore, the prob-
lem can be solved by the iterative calculation method. Procedures of
iterative calculation are described as follows:

(1) Select the idea NC command as the initial value of iterative

calculation based on the contents shown in Section 3.1, which can
be expressed by the below equation.
2 3
2 3 qwx cos C0  qwy sin C0  L sin α sin B0  q7x
x0 6 7
6 y0 7 6 qwx sin C0  qwy cos C0  L cos α sin α cos B0 7
6 7 ¼ 6 þL cos α sin α þ q7y 7
4 z0 5 6 7
4 qwz þ Lsin2 α cos B0 þ L cos α  q7z 5
1
0
2 32 3
cos C0 sin C0 0 0 xw
6 sin C0 cos C0 0 0 76 yw 7
þ4 6 7 6 7 (33)
0 0 1 0 54 zw 5
0 0 0 1 1

8
2 2
>
< cos B0 ¼ cot α  zv csc α
> cos α  xv cos α cos B0  yv sin B0 (34)
: tan C0 ¼
cos α þ xv sin B0  yv cos α cos B0

where x0 ; y0 ; z0 ; B0 ; C0 indicate the initial values of iterative calculation.

(2) Substitute ðx0 ; y0 ; z0 ; B0 ; C0 Þ into Eqs. (23)–(25), obtain the first

value of the corrected NC commandðx1 ; y1 ; z1 ; B1 ; C1 Þ, and then
assume that the i-th value of the corrected NC command
Fig. 9. The flow chart of iterative compensation methodology.
isðxi ; yi ; zi ; Bi ; Ci Þ.
(3) Substitute ðxi ; yi ; zi ; Bi ; Ci Þ into Eqs.( (28) and (30), and obtain the
0 0 0
i-th value of tool cutting trajectoryðxwi ; ywi ; zwi ; xvi0 ; yvi0 ; zvi0 Þ. where Δx ,Δy ,Δz indicate the resolution of the linear axis x,y,z in the
(4) The steps mentioned above are repeated until Eq. (35) can be machine tool respectively, and Φx ,Φy ,Φz indicate the resolution of tool
satisfied. orientation vector in x,y,z direction respectively, which are usually
8 0 defined as the resolution of the rotary axis in the machine tool.
> xwi  xw  Δx
>
> y0  yw  Δy In order to implement the iterative compensation methodology and
>
>
< z0  z  Δ
> boost the machining accuracy of the researched NOFAMT, error
wi

wi w z
(35) compensation software is developed. The flow chart displayed in Fig. 9
> x0  xv  ϕx
>
> vi explains the iterative process. The tool path can be obtained by CAD/
>
> 0
: y0vi  yv  ϕy
> CAM software based on the designed drawing of workpiece.
z  zv  ϕz
vi

91
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Table 9
Key structure parameters of DMU60P.

Work stroke Inclined angle α Resolution ðqwx ; qwy ; qwz Þ ðq7x ; q7y ; q7z Þ LðmmÞ
ðmmÞ ðmmÞ

Linear axis X 600 mm 0.001 mm

Y 700 mm
Z 600 mm
Rotary axis B 0  180∘ 45∘ 0:001∘ (-200,-200,-100)
C 0  360∘
Workpiece (100,100,200)
Tool 100

Fig. 12. Six basic geometric errors of Y- axis: (a) linear error (b) angular error.

Fig. 10. The flow chart of the verification experiment.

Fig. 13. Six basic geometric errors of Z-axis:(a) linear error (b) angular error.
Fig. 11. Six basic geometric errors of X-axis: (a) linear error (b) angular error.

92
C. Wu et al.
4. Experimental verification
4.1. Cutting experiment
To verify the effectiveness of the proposed method, an experiment is
carried out on a NOFAMT: DECKEL MAHO DMU60P, Germany. Table 9
displays its structure parameters. The experiment mainly includes three
parts: the identification of geometric errors, the acquisition of the
compensated NC code and the machining and measurement of test part.
Fig. 10 shows the flow chart of the verification experiment.
Obviously, all the geometric errors of the NOFAMT should be iden-
tified before error compensation. To obtain the 21 geometric errors of
three linear axes, the dual-frequency laser interferometer, XL-80 by
Renishaw, is utilized with nine-line method [30]. The geometric errors of
B and C axes are identified with the double bar ball, QC20-W by
Renishaw, according to the method presented by the Ref. [8]. Since the
researched NOFAMT in this paper is identical with the one in Ref. [8], the
identified results of geometric errors of rotary axes are not described
here. Note that the ambient temperature is controlled to around 20 , all
data are measured three times, and then the final error values are the
average results of three times identified values to reduce the impact of
Table 10
The value of PIGEs.(unit: rad).
εxy εxz εyz
0.000077 0.000013 0.000128
Fig. 14. Application of the S-shaped test part in cutting experiment:(a) The 3D model of the S-shaped test part (b) Tool path planning based on the 3D model of the S-shaped test part.
International Journal of Machine Tools and Manufacture 124 (2018) 80–98
thermal error and improve the measurement stability. The geometric
errors of linear axes are shown in Figs. 11–13. The PIGEs are listed
in Table 10.
The accuracy of workpiece is significantly influenced by machining
accuracy of the machine tool. Therefore, to validate the compensation
methodology described in this paper, the cutting test is a very efficient
method. Whereas, the S-shaped part is one of the most widely used
standard parts due to its additional features, which can greatly reflect the
machining accuracy of a machine tool. But, before the cutting test, tool
path, which is the prerequisite of the generation of the compensated NC
code, should be obtained. Tool path planning in this paper is carried out
by CAD/CAM Pro/E software based on the S-shaped test part designed, as
shown in Fig. 14. Then, the cutter location file (CLF) is generated by the
post process module of Pro/E software. Fig. 14(a) displays the 3D model
of the S-shaped test part. The diagram of tool path planning based on the
S-shaped test part is displayed in Fig. 14(b).
According to the identified error parameters and CLF, the compen-
sated NC codes are generated in self-developed compensation software
using the iterative compensation methodology in Section 3.
Finally, two same S-shaped test parts (as shown in Fig. 14(a)) are
machined using the studied NOFAMT with the ideal NC codes and the
εxB εBz εxC εyC
0.000015 0.000030 0.0000035 0.000060
93
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

compensated NC codes respectively. Then, the contour errors of two test

parts are measured with the coordinate measuring machine (CMM): Zeiss
CONTURA G2, Germany. As shown in Fig. 15, three different heights (i.e.
Z ¼ 3 mm, 12 mm, and 17 mm) of test part are selected to be measured,
taking the center O of the small hole as the reference coordinate system.
Each height is chosen 25 equidistant discrete points.
According to the coordinate of each measuring point, its errors of
three directions along the X, Y, and Z before and after compensation are

Table 11
The comparison of contour errors before and after compensation.

Average error (mm) Improvement (%)

Before compensation After compensation

Fig. 15. The measuring diagram for the S-shaped test part at three different heights Z ¼ 3 mm 0.0405 0.0050 87.6
(unit: mm). Z ¼ 12 mm 0.0474 0.0074 84.4
Z ¼ 17 mm 0.0535 0.0181 81.2

Fig. 16. The contour error of measured points at Z ¼ 3 mm.

Fig. 17. The contour error of measured points at Z ¼ 12 mm.

Fig. 18. The contour error of measured points at Z ¼ 17 mm.

94
C. Wu et al.
Fig. 19. Flow diagrams for software error compensation method: (a) the iterative or differential method; (b) the algebraic method; (c) the proposed method.
calculated on the basis of Eq. (32) respectively. Thus, the contour error of
the corresponding point before and after compensation is calculated.
Then, the contour errors of 75 points for the test part before and after
compensation are all obtained, as shown in Figs. 16–18. The comparison
of the test part before and after compensation is shown in Table 11. It can
be seen that the contour errors of all 75 points after compensation are
much smaller than that before compensation. The maximum error is
decreased from 0.0532 mm to 0.0213 mm, from 0.0639 mm to
0.0206 mm, and from 0.0718 mm to 0.0275 mm at Z ¼ 3 mm, 12 mm
and17 mm respectively. Note that the maximum error before and after
compensation is located at the measured point 23, 25 and 25 at
Z ¼ 3 mm, 12 mm and17 mm respectively. While the average error is
reduced from 0.045 mm to 0.0050 mm, from 0.0474 mm to 0.0074 mm,
and from 0.0535 mm to 0.0181 mm at Z ¼ 3 mm, 12 mm and17 mm
respectively. In other words, the accuracy of test part after compensation
is improved about 87.6%, 84.4% and 81.2% at Z ¼ 3 mm, 12 mm and
17 mm respectively. The average improvement after compensation is
about 84.2%. Therefore, it is obvious that the proposed iterative
compensation methodology based on TRMCE is precise and effective to
improve machining accuracy of non-orthogonal multi-axis ma-
chine tools.
4.2. Expanding discussion of the new iterative method
At present, the commonly used methods of software compensation
mainly include iterative method, differential method and algebraic
method, as shown in Table 1. In these methods, there are two ways to
modify the NC codes, one is to superimpose additional codes on the basis
of the ideal NC code, such as the iterative method, differential method,
and the other is to obtain the modified NC codes directly considering the
geometric errors of machine tools, such as algebraic method. They pay
much more attention to how to obtain the modified NC codes of tool path
without concerning the real effect of the obtained NC codes on tool
cutting trajectory. Obviously, the new iterative method takes the differ-
ence between tool path and tool cutting trajectory as the control target
and can directly obtain the actual NC code, which can directly drive the
machine tool to achieve the desired machining accuracy. What's more, it
does not have the inverse calculation of traditional methods. Fig. 19
displays contrastive flow diagrams for the core ideas of these compen-
sation methods. Therefore, the proposed method in this paper is more
direct and accurate.
It is worth noting that different topological structures of a machine
tool can be used to establish different TRMCEs, which leads to the
different mathematical expressions of the iterative calculation method.
Thanks for the MBS theory, which is a systematic and universal modeling
method; the iterative calculation procedures of the actual NC command
95
International Journal of Machine Tools and Manufacture 124 (2018) 80–98
are suitable for other type of machine tools, especially for more than
three-axis precision or ultra-precision machine tools. Hence, for any
machine tool structure, the core idea of the method can not only be
applied to compensate the geometric error conveniently and effectively,
but also thermal error, load error, and so on. Furthermore, with the
change of the inclined angle α for the NOFAMT, the proposed method can
be straightforwardly suitable for any type of machine tools with the same
topological structure.
5. Conclusion
This paper proposes a new iterative compensation method of geo-
metric errors to improve the machining accuracy of a non-orthogonal
five-axis machine tool (NOFAMT). Compared with previous methods
[3,5,12–18], The new approach has the following advantages: (1) it takes
the difference between tool path and tool cutting trajectory as the control
objective and can directly obtain the actual numerical control (NC) codes
controlling the machine tool to achieve the desired machining accuracy,
(2) the real response of the machine tool on the NC codes obtained are
concerned and (3) it does not have the inverse calculation of tradi-
tional methods.
To verify the effectiveness of the proposed method, a cutting exper-
iment is carried out. Experimental results show the machining accuracy
of an S-shaped test part with geometric error compensation is improved
about 87.6%, 84.4% and 81.2% at Z ¼ 3 mm, 12 mm and 17 mm
respectively, compared with that without compensation. Furthermore,
the average machining accuracy after compensation is improved about
84.2%. Therefore, there is no doubt that the method is accurate and
effective for machining accuracy improvement. Similarly, if other error
sources are considered, the method can be executed as well. Further-
more, the proposed method in this paper can be straightforwardly suit-
able for any type of machine tools with the same topological structure.
Despite the progress made in this work, only geometric errors are
considered. However, some other errors, such as thermally induced error,
cutting-force-induced error, dynamic load-induced error etc., also affect
the machining accuracy of machine tools. Therefore, further researches
should be done for all of the errors, or at least most of them.
Acknowledgements
This work is financially supported by the National Natural Science
Foundation of China (No. 51775010 and 51705011), Science and
Technology Major Projects of High-end CNC Machine Tools and Basic
Manufacturing Equipment of China (No.2014ZX04011031). The authors
appreciate the support from Key Scientific Research Project for Henan
Province Higher school of China (No.16B460015 and 18A460036).
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Appendix E. Supplementary data

Supplementary data related to this article can be found at https://doi.org/10.1016/j.ijmachtools.2017.07.008.

Appendix A
8 E ¼ δ ðxÞ þ δ ðzÞ þ q þ x þ zðε ðxÞ þ ε Þ þ δ ðBÞcosB þ δ ðBÞsin B þ q ðε ðxÞ þ ε ðzÞ þ ε Þ
>
px x x 7x y xz x y 7z y y xz

>
> q ðε ðxÞ þ ε Þ þ Lðsin α sin B  ðcos α þ sin αcosBÞðε ðxÞ þ ε ðzÞ þ ε Þ  cosαðcosBε ðBÞ
7y
2 2

>
z xy y y xz y

>
>
þsin Bε ðBÞÞ  cosα sin αðε ðxÞ þ ε ðzÞ þ ε þ ε Þ þ sin αcosBε ðBÞ þ cosα sin αcosBðε ðxÞ
x z z xB xy z z

>
>
þε ðzÞ þ ε þ ε ÞÞ þ ðq þ y Þðsin C þ cos Cε ðCÞ þ cos Cε ðyÞÞ  ðq þ x Þðcos C  sin Cε ðCÞ  sin C
z xB xy wy w z z wx w z
>
> ε ðyÞÞ  ðq þ z Þðε ðCÞ þ ε ðyÞ þ cos Cε ðCÞ þ sin Cε ðCÞÞ  cos Cδ ðCÞ þ sin Cδ ðCÞ  δ ðyÞ
>
z wz w x y y x x y x

< Esin αδ
¼ δ ðxÞ þ δ ðzÞ þ q þ xε  zðε ðxÞ þ ε Þ  q ðε ðxÞ þ ε ðzÞ þ ε Þ þ q ðε ðxÞ þ ε ðzÞ þ ε Þ
py y y 7y
ðBÞ þ cosαðcosBδ ðBÞ þ sin Bδ ðBÞÞ þ Lðsin αε ðBÞ þ cosα sin α þ cos αðε ðxÞ þ ε ðzÞ
xy x yz 7z
2
x x yz 7x z
2
z xy
z y x x x x
þεBz þ εyz þ cosBεx ðBÞ  sin Bεy ðBÞÞ þ sin2 αcosBðεx ðxÞ þ εx ðzÞ þ εyz þ εBz Þ þ sin α sin Bðεz ðxÞ (A1)
>
> þε ðzÞ þ ε þ ε ÞÞ  δ ðyÞ  y  ðq þ x Þðsin C þ cos Cε ðCÞ þ cos Cε ðyÞÞ  ðq þ y Þðcos C  sin C
>
>
z xy xB x
ε ðCÞ  sin Cε ðyÞÞ þ ðq þ z Þðε ðyÞ þ ε ðCÞ þ cos Cδ ðCÞ  sin Cδ ðCÞÞ  cos Cδ ðCÞ  sin Cδ ðCÞ
wx w z z wy w

>
>
z z wz w x y x y y x

>
>
E ¼ δ ðxÞ þ δ ðzÞ þ q þ z þ cosαδ ðBÞ þ q ðε ðxÞ þ ε ðzÞ þ ε Þ  q ðε ðxÞ þ ε ðzÞ þ ε Þ þ sin α
pz z z 7z z 7y x x yz 7x y y xz

>
> ðcosBδ ðBÞ þ sin Bδ ðBÞÞ þ Lðcos α  sin αcosB þ cosα sin αðε  ε ðBÞ þ ε Þ  sin α sin B 2 2

>
y x Bz x yz

: ðε ðxÞ þ ε ðzÞ þ ε Þ þ cosα sin αcosBðε  ε ðxÞ  ε ðzÞ  ε  ε Þ  cosα sin α sin Bε ðBÞÞ
y y xz
δz ðCÞ  δz ðyÞ  qwz  zw þ ðqwx þ xw Þðεy ðCÞ þ cos Cεx ðCÞ þ cos Cεy ðyÞ þ sin Cεx ðyÞ þ sin Cεy ðCÞÞ
xB x x Bz yz y

ðqwy þ yw Þðεx ðCÞ þ cos Cεx ðyÞ þ cos Cεy ðCÞ þ sin Cεx ðCÞ þ sin Cεy ðyÞÞ

8 Evx ¼ sin Bðsinα  cos αεx ðBÞÞ þ cos Bðsinαεz ðBÞ  cos αεy ðBÞ  sin2 αðεy ðxÞ þ εy ðzÞ þ εxz Þ
>> þcos αsinαðεz ðxÞ þ εz ðzÞ þ εxB þ εxy ÞÞ  cos2 αðεy ðxÞ þ εy ðzÞ þ εxz Þ  cos αsinαðεz ðxÞ

< þεz ðzÞ þ εxB þ εxy Þ  xv ðcosC  sin Cεz ðCÞ  sin Cεz ðyÞÞ  zv ðεx ðCÞ þ εy ðyÞ þ cosC
εy ðCÞ þ sin Cεx ðCÞÞ þ yv ðsin C þ cosCεz ðCÞ þ cosCεz ðyÞÞ
Evy ¼ cos2 αðεx ðxÞ þ εx ðzÞ þ εBz þ εyz Þ þ sin2 αεx ðBÞ þ cos αsinα þ cos Bðcos αsinαþ
cos2 αεx ðBÞ þ sin2 αðεx ðxÞ þ εx ðzÞ þ εBz þ εyz ÞÞ þ sin Bðsinαðεz ðxÞ þ εz ðzÞ þ εxB þ

>>
(A2)
εxy Þ  cos2 αεy ðBÞ þ cos αsinαεz ðBÞÞ  yv ðcosC  sin Cεz ðCÞ  sin Cεz ðyÞÞ þ zv ð

:
εx ðyÞ þ εy ðCÞ þ cosCεx ðCÞ  sin Cεy ðCÞÞ  xv ðsin C þ cosCεz ðCÞ þ cosCεz ðyÞÞ
Evz ¼ cos αsinαðεx ðxÞ  εx ðBÞ þ εx ðzÞ þ εBz þ εyz Þ  cos2 α  cos Bðsin2 α þ cos αsinα
ðεx ðxÞ  εx ðBÞ þ εx ðzÞ þ εBz þ εyz ÞÞ  sin Bsinαðεy ðxÞ þ εy ðzÞ þ εxz  sinαεz ðBÞ
þcos αεy ðBÞÞ  zv þ xv ðεy ðCÞ þ cosCεx ðCÞ þ cosCεy ðyÞ  sin Cεx ðyÞ  sin Cεy ðCÞÞ
yv ðεx ðCÞ þ εx ðyÞ þ cosCεy ðCÞ þ sin Cεx ðCÞ þ sin Cεy ðyÞÞ

Appendix B
8 temp1 ¼ x ðsin αðε ðxÞ  ε ðCÞ  ε ðyÞ þ ε ðzÞ þ ε þ ε Þ  cos αsin αðε ðCÞ
5 4

>
>
v z
ε ðxÞ þ ε ðyÞ  ε ðzÞ  ε ÞÞ þ y cos αsin αð3ε ðBÞ  2ε ðxÞ  2ε ðzÞþ
x z
4
z xB xy x

>
>
y
ε ðCÞ  2ε  2ε Þ þ z sin αε ðCÞ;
y y xy
5
v x x x

>
>
y Bz
5
xy
temp2 ¼ x ðsin αðε ðxÞ  1 þ ε ðzÞ þ ε þ ε Þ þ cos αsin αð3ε ðBÞ  2ε ðxÞ þ ε
v x
4

< ðyÞ  2ε ðzÞ þ ε ðCÞ þ ε ðxÞ þ ε ðzÞ  2ε þ ε  2ε ÞÞ þ y ðsin αðε ðCÞ

x
v
y
4
z
y
z
y
xB
Bz
xy

5
xy xz v
x
5
z
x x

þεz ðyÞ þ cos αsin αðεx ðCÞ þ εy ðyÞÞÞ  zv sin αεy ðCÞ; Þ
temp3 ¼ zv ðsin5 αðεx ðxÞ þ εx ðyÞ þ εx ðzÞ þ εy ðCÞ þ εBz þ εyz Þ  cos αsin4 αÞ þ cos α (B1)
>
> sin6 α þ εx ðBÞðsin7 α  4cos2 αsin5 α þ 3cos4 αsin3 αÞ þ ðεBz þ εyz Þð5cos2 α
>
> sin5 α  cos4 αsin3 αÞ  cos3 αsin4 α;
>
> temp4 ¼ y3v sin2 αεy ðCÞ  x3v sin2 αðεy ðyÞ þ εx ðCÞÞ  sin2 αðxv y2v ðεx ðCÞ þ εy ðyÞÞ
: x2v yv εy ðCÞÞ;
temp5 ¼ ðx3v ðεx ðCÞ þ εy ðyÞÞ  y3v εy ðCÞ þ xv y2v ðεx ðCÞ þ εy ðyÞÞ  x2v yv εy ðCÞÞðcos2 α
þzv Þ:

96
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Appendix C
8
> u1 ¼ sinαεz ðBÞ  cos αεy ðBÞ þ sin2 αðεx ðCÞ  εy ðxÞ þ εy ðyÞ  εy ðzÞ  εxz Þ;
>
>
>
> u2 ¼ ðsin αðεz ðCÞ þ εz ðxÞ  εz ðyÞ þ εz ðzÞ þ εxB þ εxy Þ þ cos2 αεy ðBÞ  cos α sin αεz ðBÞÞ;
>
> 2 2
> u3 ¼ cos αεx ðBÞ þ sin αðεx ðxÞ  εx ðyÞ þ εx ðzÞ  εy ðCÞ þ εBz þ εyz Þ  cos α sin αð1 þ εz ðCÞ
>
>
>
> εz ðxÞ þ εz ðyÞ  εz ðzÞ  εxB  εxy Þ;
> u4 ¼ cos2 αεx ðBÞ  cos α sin α þ sin2 αðεx ðxÞ  εx ðyÞ þ εx ðzÞ  εy ðCÞ þ εBz þ εyz Þ;
>
>
>
< u5 ¼ sin αðεz ðxÞ  εz ðCÞ  εz ðyÞ þ εz ðzÞ þ εxB þ εxy Þ  cos2 αεy ðBÞ;
v ¼ cos2 αðε ðCÞ  ε ðxÞ þ ε ðyÞ  ε ðzÞ  ε Þ þ cos α sin αðε ðCÞ  ε ðxÞ þ ε ðyÞ
1 (C1)
>
>
x
εz ðzÞ  εxB  εxy Þ;
y y y xz z z z
>
>
>
> v2 ¼ cos αsinα þ cos2 αðεx ðxÞ  εx ðyÞ þ εx ðzÞ  εy ðCÞ þ εBz þ εyz Þ þ sin2 αεx ðBÞ;
>
>
>
> v3 ¼ cos α sin α þ cos2 αðεx ðxÞ  εx ðyÞ þ εx ðzÞ  εy ðCÞ þ εBz þ εyz Þ þ sin2 αεx ðBÞ;
>
> v4 ¼ ðcos2 αðεx ðCÞ þ εy ðyÞÞ þ cos α sin αðεz ðCÞ þ εz ðyÞÞÞ;
>
>
>
> 2
: v5 ¼ cos α sin αðεx ðBÞ  εx ðxÞ  εx ðzÞ  εBz  εyz þ εx ðyÞ þ εy ðCÞÞ  sin α;
v6 ¼ ðsin αðεy ðxÞ þ εy ðzÞ þ εxz  εx ðCÞ  εy ðyÞÞ þ sin2 αεz ðBÞ þ cos α sin αεy ðBÞÞ:

8
> ux ¼ q7x ðcosC  sin Cðεz ðCÞ  εz ðxÞ þ εz ðyÞ  εz ðzÞ  εxy ÞÞ þ q7y ðsin C þ cosCðεz ðCÞ
>
>
>
> εz ðxÞ þ εz ðyÞ  εz ðzÞ  εxy ÞÞ þ q7z ðcosCðεy ðxÞ  εx ðCÞ  εy ðyÞ þ εy ðzÞ þ εxz  εx
>
>
> ðxÞÞ þ sin Cðεx ðyÞ  εx ðzÞ þ εy ðCÞ  εyz Þ  εy ðCÞÞ  qwx  δx ðCÞ þ cosCðδx ðxÞ  δx
>
>
>
>
> ðyÞ þ δx ðzÞÞ þ sin Cðδy ðxÞ  δy ðyÞ þ δy ðzÞÞ þ cos B cos Cδx ðBÞ þ cos α cos B sin C
>
>
> 2
> y ðBÞ þ cosα sin B sin Cδx ðBÞ þ Lðcosα sin α sin C þ cos C sin α sin B þ cos αðεy
>
>
δ
>
> ðCÞ þ cos B sin Cε ðBÞ  sin B sin Cε ðBÞ þ cosCðε ðCÞ  ε ðxÞ þ ε ðyÞ  ε ðzÞ
> x y x y y
> ε Þ þ sin Cðε ðxÞ  ε ðyÞ þ ε ðzÞ  ε ðCÞ þ ε þ ε ÞÞ þ sin2 αðsin Cε ðBÞ þ cos B
y
>
>
>
>
xz x x x y Bz yz x
> εy ðCÞ þ cos BcosCðεx ðCÞ  εy ðxÞ þ εy ðyÞ  εy ðzÞ  εxz Þ þ cos B sin Cðεx ðxÞ  εx ð
>
>
>
>
> yÞ þ εx ðzÞ  εy ðCÞ þ εBz þ εyz ÞÞ  cos α cos B cos Cεy ðBÞ  cos α cos C sin Bεx ðBÞþ
>
>
>
> cos α cos C sin αðεz ðCÞ  εz ðxÞ þ εz ðyÞ  εz ðzÞ  εxB  εxy Þ þ cos BcosC sin αεz ðBÞ
>
>
>
> þcos α cos B sin α sin C  sin α sin B sin Cðεz ðCÞ  εz ðxÞ þ εz ðyÞ þ εz ðzÞ þ εxB þ εxy
>
>
>
> Þcos α cos BcosCðεz ðCÞ þ εz ðxÞ þ εz ðyÞ þ εz ðzÞ þ εxB þ εxy ÞÞ;
>
>
>
> uy ¼ q7x ðsin C  cosCðεz ðCÞ  εz ðxÞ þ εz ðyÞ  εz ðzÞ  εxy ÞÞ þ q7y ðcosC þ sin Cðεz ð
>
>
>
>
> xÞ þ εz ðzÞ þ εxy  εxz Þ þ εx ðCÞÞ þ q7z ðcosCðεx ðyÞ  εx ðxÞ  εx ðzÞ þ εy ðCÞ þ εyz Þþ
>
>
> sin Cðεx ðCÞ  εy ðCÞ  εy ðyÞ  εy ðzÞÞ  qwx sin2 Cðεz ðCÞ þ εz ðyÞÞ  qwy ð1 þ cos C
>
>
>
>
< sin Cðεz ðCÞ þ εz ðyÞÞÞ þ cosCδy ðxÞ  δy ðCÞ  cosCðδy ðyÞ  δy ðzÞÞ  sin Cðδx ðxÞ
δx ðyÞ þ δx ðzÞÞ  cos B sin Cδx ðBÞ  cos C sin αδz ðBÞ þ sin B sin Cδy ðBÞ þ cosα (C2)
>
>
> cos B cos Cδy ðBÞ þ cosα sin B cos Cδx ðBÞ þ Lðþcosα sin α cos C  sin α sin B sin
>
>
> 2
> C þ cos αðcos Cðεx ðxÞ  εx ðyÞ þ εx ðzÞ  εy ðCÞ þ εBz þ εyz Þ  sin Cðεx ðCÞ  2εy ðxÞ
>
>
>
> þεy ðyÞ  εy ðzÞ  εxz Þ þ cos B cos Cεx ðBÞ  εx ðCÞ  sin B cos Cεy ðBÞÞ þ sin αðcos
>
>
>
> Cεx ðBÞ  cos B cos Cðεx ðyÞ  εx ðzÞ þ εy ðCÞ  εBz  εyz Þ  cos B sin Cðεx ðCÞ  εy ðx
>
>
>
> Þ þ εy ðyÞ  εy ðzÞ  εxz Þ  cos Bεx ðCÞÞ þ cosα sin α cos B cos C þ cosα cos B sin Cεy
>
>
>
>
> ðBÞ þ cosα sin B sin Cεx ðBÞ  cos B sin α sin Cεz ðBÞ  sin α sin B cos Cðεz ðCÞ  εz
>
>
> ðxÞ þ εz ðyÞ  εz ðzÞ  εxB  εxy Þ þ ðcosα sin α sin C  cosα sin α cos B sin CÞðεz ðxÞ
>
>
>
> þεz ðzÞ þ εxB þ εxy ÞÞ;
>
>
>
>
> uz ¼ q7x ðcosCεy ðCÞ þ sin Cεx ðCÞ þ εx ðCÞ  εy ðxÞ þ εy ðyÞ  εy ðzÞ  εxz Þ þ q7y ðsin Cεy
>
>
>
> ðCÞ  cosCεx ðCÞ þ εx ðxÞ  εx ðyÞ þ εx ðzÞ  εy ðCÞ þ εyz Þ þ q7z  qwz þ δz ðxÞ  δz ðC
>
>
>
> Þ  δz ðyÞ þ δz ðzÞ þ cosαδz ðBÞ þ sinαsinBδx ðBÞ þ cos B sin αδy ðBÞ  Lðcos2 αþ
>
>
> sin2 αcosB þ cos α sin αðεx ðBÞ  εx ðxÞ þ εx ðyÞ  εx ðzÞ þ εy ðCÞÞ þ sin αsinBðεy ð
>
>
>
>
> xÞ  εx ðCÞ  εy ðyÞ þ εy ðzÞÞ  cos α sin αsin2 CðεBz þ εyz Þ  sin αsinBsin2 Cεxz þ
>
>
>
> cos α sin αcosCεx ðCÞ  ðcos α sin α sin C þ cos C sin α sin BÞεy ðCÞ  ðsin α sin B
>
>
>
>
: sin C  cos α sin α cos BcosCÞεx ðCÞ  cos α sin α cos B sin Cεy ðCÞ þ cos α sin α
>
cos Bðεx ðxÞ  εx ðBÞ þ εx ðyÞ þ εx ðzÞ þ εy ðCÞ þ εBz þ εyz Þ þ cos α sin α sin Bεy ðBÞÞ:

97
C. Wu et al. International Journal of Machine Tools and Manufacture 124 (2018) 80–98

Appendix D
2 sin Cðεx ðyÞ þ εy ðCÞ
3
cos C  sin Cðεz ðCÞ sin C  cos C εx ðxÞ  εyz Þ  cosC
6 þε 07
6 z ðyÞ  εxy Þ ðεz ðCÞ þ εz ðyÞÞ ðεx ðCÞ þ εy ðyÞ þ εy ðxÞ 7
6 þεxz Þ  εy ðCÞ 7
6 cosCðεx ðyÞ þ εy ðCÞ 7
Tw ¼ 6 sin C  cos Cðεz ðCÞ 7
6 þε ðyÞ  ε Þ cos C þ sin C εx ðxÞ  εyz Þ þ sin C
07
6 z ðεz ðCÞ þ εz ðyÞÞ ðεx ðCÞ þ εy ðyÞ þ εy ðxÞ 7
6 7
xy
þεxz Þ þ εx ðCÞ
4 εx ðCÞ þ εy ðyÞ þ cos C εy ðCÞ þ εx ðyÞ þ cos C 5
1 0
εy ðCÞ þ sin Cεx ðCÞ εx ðCÞ  sin Cεy ðCÞ
2 0 0 0 3 1
u1 sinα  cos αεx ðBÞ u2 u3
6 u4 u5 0 ðεx ðCÞ þ εy ðyÞÞ 7
Tv1 ¼ 6
4 cos α sin αεx ðCÞ
7
sin αεy ðCÞ sin αεx ðCÞ cos α sin αεy ðCÞ 5
2 0 0 0 3 0
v1 v2 sin2 αεy ðCÞ 0
6 v3 v4 εx ðCÞ 07
Tv2 ¼ 6
4 cos α sin αεx ðCÞ cos α sin αεy ðCÞ
7
v5 v6 5
0 0 0 0

References [16] Jianxiong Chen, Shuwen Lin, Bingwei He, Geometric error compensation for multi-
axis CNC machines based on differential transformation, Int. J. Adv. Manuf.
Technol. 71 (2014) 635–642.
[1] Y. Ma, Sensor Placement Optimization for Thermal Error Compensation on Machine
[17] Shuang Ding, Xiaodiao Huang, Chunjian Yu, et al., Actual inverse kinematics for
Tools [D], University of Michigan, Ann Arbor, 2001.
position-independent and position-dependent geometric error compensation of
[2] M.S. Uddin, S. Ibaraki, A. Matsubara, et al., Prediction and compensation of
five-axis machine tools, Int. J. Mach. Tools Manuf. 111 (2016) 55–62.
machining geometric errors of five-axis machining centers with kinematic errors,
[18] Xiangdong Zhou, Zhouxiang Jiang, Bao Song, et al., A compensation method for the
Prec Eng. 33 (2009) 194–201.
geometric errors of five-axis machine tools based on the topology relation between
[3] Guoqiang Fu, Jianzhong Fu, Hongyao Shen, et al., Product-of-exponential formulas
axes, Int. J. Adv. Manuf. Technol. 88 (2017) 1993–2007.
for precision enhancement of five-axis machine tools via geometric error modeling
[19] Qiang Cheng, Hongwei Zhao, Guojun Zhang, et al., An analytical approach for
and compensation, Int. J. Adv. Manuf. Technol. 81 (2015) 289–305.
crucial geometric errors identification of multi-axis machine tool based on global
[4] Guoda Chen, Yingchun Liang, Yazhou Sun, et al., Volumetric error modeling and
sensitivity analysis, Int. J. Adv. Manuf. Technol. 75 (2014) 107–121.
sensitivity analysis for designing a five-axis ultra-precision machine tool, Int. J. Adv.
[20] K.F. Eman, B.T. Wu, M.F. DeVries, A generalized geometric error model for multi-
Manuf. Technol. 68 (2013) 2525–2534.
axis machines, CIRP Ann. 36 (1987) 253–256.
[5] Nuodi Huang, Yongqiao Jin, Qingzhen Bi, Integrated post-processor for 5-axis
[21] Binit Kumar Jha, Anjani Kumar, Analysis of geometric errors associated with five-
machine tools with geometric errors compensation, Int. J. Mach. Tools Manuf. 94
axis machining centre in improving the quality of cam profile, Int. J. Mach. Tools
(2015) 65–73.
Manuf. 43 (2003) 629–636.
[6] M.A. Mannan, R. Ramesh, A.N. Poo, Error compensation in machine tools-a review.
[22] Psang Dain Lin, Chian Sheng Tzeng, Modeling and measurement of active
Part I: geometric, cutting-force induced and fixture-dependent errors, Int. J. Mach.
parameters and workpiece home position of a multi-axis machine tool, Int. J. Mach.
Tool. Manuf. 40 (2000) 1235–1256.
Tools Manuf. 48 (2008) 338–349.
[7] Samir Mekid, Tunde Ogedengbe, A review of machine tool accuracy enhancement
[23] J.W. Fan, J.L. Guan, W.C. Wang, et al., A universal modeling method for
through error compensation in serial and parallel kinematic machines, Int. J. Precis.
enhancement the volumetric accuracy of CNC machine tools, J. Mater. Process.
Technol. 1 (2010) 251–285.
Technol. 129 (2002) 624–628.
[8] Dongju Chen, Lihua Dong, Yanhua Bian, et al., Prediction and identification of
[24] Jixiang Yang, J.R.R. Mayer, Yusuf Altintas, A position independent geometric errors
rotary axes error of non-orthogonal five-axis machine tool, Int. J. Mach. Tools
identification and correction method for five–axis serial machines based on screw
Manuf. 94 (2015) 74–87.
theory, Int. J. Mach. Tools Manuf. 95 (2015) 52–66.
[9] T. Moriwaki, Multi-functional machine tool, CIRP Ann. - Manuf. Technol. 57 (2008)
[25] Shijie Guo, Dongsheng Zhang, Yang Xi, Global quantitative sensitivity analysis and
736–749.
compensation of geometric errors of CNC machine tool, Math. Problems Eng.
[10] Chenhua She, Zhaotang Huang, Postprocessor development of a five-axis machine
(2016) 1–12.
tool with nutating head and table configuration, Int. J. Adv. Manuf. Technol. 38
[26] Qiang Cheng, Ziling Zhang, Guojun Zhang, et al., Geometric accuracy allocation for
(2008) 728–740.
multi-axis CNC machine tools based on sensitivity analysis and reliability theory,
[11] Chenhua She, Chuncheng Chang, Development of a five-axis postprocessor system
Proc IMechE Part C J Mech. Eng. Sci. 229 (2015) 1134–1149.
with a nutating head, J. Mater. Process. Technol. (2007) 60–64, 187–188.
[27] K.I. Lee, S.H. Yang, Robust measurement method and uncertainty analysis for
[12] Gangwei Cui, Yong Lu, Jianguang Li, et al., Geometric error compensation software
position-independent geometric errors of a rotary axis using a double ball-bar, Int.
system for CNC machine tools based on NC program reconstructing, Int. J. Adv.
J. Precis. Eng. Manuf. 14 (2013) 231–239.
Manuf. Technol. 63 (2012) 169–180.
[28] Nuodi Huang, Yongqiao Jin, Qingzhen Bi, et al., Integrated post-processor for 5-axis
[13] Shuang Ding, Xiaodiao Huang, Chunjian Yu, et al., Novel method for position-
machine tools with geometric errors compensation, Int. J. Mach. Tools Manuf. 94
independent geometric error compensation of five-axis orthogonal machine tool
(2015) 65–73.
based on error motion, Int. J. Adv. Manuf. Technol. 83 (2016) 1069–1078.
[29] Huimin Gao, Fengzhou Fang, Xiaodong Zhang, Reverse analysis on the geometric
[14] Abdul Wahid Khan, Wuyi Chen, A methodology for systematic geometric error
errors of ultra-precision machine, Int. J. Adv. Manuf. Technol. 73 (2014)
compensation in five-axis machine tools, Int. J. Adv. Manuf. Technol. 53 (2011)
1615–1624.
615–628.
[30] Y. Liu, L. Liu, X. Zhao, et al., Investigation of error compensation technology for NC
[15] Shaowei Zhu, Guofu Ding, Shengfeng Qin, et al., Integrated geometric error
machine tool, China, Mech. Eng. 12 (1998) 48–51 (in Chinese).
modeling, identification and compensation of CNC machine tools, Int. J. Mach.
Tools Manuf. 52 (2012) 24–29.

98