A Force Monitoring Method Based on Built-in Sensors for Feed System of Computer Numerical Control Machine Tools
TELKOMNIKA, Vol.14, No.3A, September 2016, pp. 345~352
ISSN: 1693-6930, accredited A by DIKTI, Decree No: 58/DIKTI/Kep/2013
DOI: 10.12928/TELKOMNIKA.v14i3A.4429
345
A Force Monitoring Method Based on Built-in Sensors
for Feed System of Computer Numerical Control
Machine Tools
Guangsheng Chen*, Qingzhen Zheng
School of Mechanical Engineering, University of Shanghai for Science and Technology, Shanghai City,
200093, P. R. China
*Corresponding author, e-mail: cgs-168@163.com
Abstract
As force has important influence to the performance and machining quality of machine tool during
machining process, this paper proposed a method to monitor the load force of feed system using build-in
sensor of the machine tools. To reduce the influence of noise caused by external disturbance and
difference operation, a state observer based on Kalman filter that optimized values of force, velocity and
displacement of feed system is adopted, and gain coefficient of the observer is obtained by recursive
algorithm. To validate the effectiveness of the method, friction force testing experiments under no-load and
force testing experiments under complex cutting conditions are performed respectively, the results show
that data obtained by the method are effective, and this method significantly restrained the external noise
and noise caused by difference operation. The method suits to on-line and real-time force monitoring in
production without additional sensors.
Keywords: Kalman filter, feed system, force monitoring, built-in sensor
Copyright © 2016 Universitas Ahmad Dahlan. All rights reserved.
1. Introduction
Large CNC machine tools are key equipment in civil and national defense industry for
machining precision parts which are complex, large-scale, thin-walled and hard to machine.
High speed and high precision are the important development directions of CNC machines.
However, accuracy and stability problems arise as the increasing of the rotating speed of the
spindle and the feeding speed of servo system. High feed rate means the enlargement of the
system tracking error. To reduce the tracking error and improve the machining accuracy of
machine tools, researchers have taken some measures on the controller of machine tool, such
as improving the position and speed gains of the machine controller and using the feed forwards
of speed or acceleration etc. [1, 2]. However, the high-gain parameters affect the stability of the
control system. In addition, the stability problem of high-speed cutting, i.e. the cutting
instability(such as chatter) in the processing process in the case of large cut depth, high rotation
speed of the spindle, and high feeding speed condition, often plague the researchers and users.
On the other hand, when a large CNC machine tools work under unconventional conditions
such as high velocity, high acceleration, large load, large displacement, forces(including cutting
force, sliding friction ) of feed system are likely to cause vibration, impact, deformation, and
wear of components during machining ,thereby, processing system, spindle system and feed
system of CNC are significantly influenced. To meet the requirements including high precision,
high efficiency, and high performances of service as well as being intelligent under poor
machining condition, it is necessary to monitor load force on-line and real-time. The forces in
processing system can be obtained using tri-axial force sensor [3], however, as an external
sensor, which is often used in experimental studies and difficult to be applied in industrial
conditions because of inconvenience in installation. ZHOU Yuqinq [4] and RE Haber [5] et al
proposed that force conditions can be indirectly reflected by current signal of servo motor of the
machine, then requirements for performance evaluation and intelligent control of large CNC
machine tool can be achieved by monitoring and detecting the forces of feed system. However,
this method is only applied in some detection occasions due to the constraint of monitor
principle.
Received April 9, 2016; Revised July 24, 2016; Accepted August 6, 2016
346
ISSN: 1693-6930
This paper proposes a state observer based on Kalman filter. The observer can realize
the optimal estimation of feed system force state by collecting the built-in sensor signals of feed
system, such as motor current and motor encoder (or grating ruler) displacement etc.. Thereby,
the various states (no-load, online machining etc.) of CNC can be monitored conveniently.
2. The Monitoring Principle of Force in Feed System Based on Motor Current Signal
Typical feed system of CNC machine tool is comprised by amplifier, servo motor, ball
screw, screw nut, slide guide and worktable. Ball screw is driven by motors to make screw nut
move. Thereby, the rotating motion of the motors is transformed into linear motion. Worktable is
supported by sliding or rolling slide guide. And servo motor and ball-screw are directly
connected by couplings. To simplify the control model, the elasticity of the worktable is not
considered, and the mass and damping of all moving parts are equivalent to the equivalent
inertia and equivalent damping of motor shaft. The simplified electromechanical coupling control
system model of the feed system is showed in Figure 1 [6]. And the symbols in Figure 1 are
defined as the follows: u: the input signal of the controller, i: the current of the AC servo motor,
ym: the equivalent displacement of screw nut, l: screw lead, Kbs: the transferred coefficient of
the nut displacement corresponding to motor rotation angle, θ: the rotation angle of motor, ω :
the rotation velocity of motor, Kpp: the proportional gain of position loop, Kvp: the gain of
velocity loop, Kcp: the gain of current loop, Kt : motor torque constant, L and R: equivalent
inductance and equivalent resistance of motor respectively, s: Laplace transform operator, Je :
equivalent inertia, Bm: equivalent damping.
Figure 1. Simplified electromechanical control system of feed system
For the motor shaft, the equation is:
g
iK t = J e ω+ Be ω + Te
in the equation (1),
working state:
Te
(1)
is the equivalent torque of external load to the motor shaft, in
Te = T f + Tc
Tf is the friction in feed system, and Tc is the cutting force of the feed system in the feed
direction, so equation (1) equals to:
Te = iKt - J e ω- Be ω
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
(2)
TELKOMNIKA
ISSN: 1693-6930
347
If Kt, Je, Be, i, ω , ω were known, forces of the feed system (including external cutting
force and friction force) can be obtained by equation (2). In this equation, Kt can be obtained
from to motor manual, Jeand Be can be acquired by identification or calculation. However, ω and
ω
in the equation can not be obtained directly. To obtain the two values, usually, we should
firstly get the motor displacement sensor (encoder or grating ruler) signal y. By the difference
operation of y, ω is obtained. Then through the differential operation of ω , ω is acquired.
However, due to the error amplification of difference operation, especially in the real-time
monitoring system with sampling time on millisecond level, the signal-to-noise ratio of ω and ω
is sharply weakened. Especially after 2 times of difference operation, signal ω is almost
covered by calculation noise. Thus the detected force signal is provided with very low signal-tonoise ratio which even can be meaningless for research. To accurately monitoring the external
forces of feed system, a state observer based on Kalman filter is designed for conducting
optimal value observation to motor rotation speed and external torque.
3. Design of Kalman Filter
Kalman filter algorithm is proposed by American scholar Rudolf. Kalman. It is an optimal
forecasting estimation method based on minimum variance and can provide solutions to the
random noise disturbance in measured signal. Equation (1) can be written as discrete state
equation [7] which includes external disturbances, measurement and input noise.
x ( k 1)
x(k )
( k 1) A ( k ) B[u ( k )] W u ( k )
wd ( k )
d ( k )
d ( k 1)
x(k )
m ( k )
( k ) V
(k )
C
d m (k )
d ( k )
(3)
d (k ) is the force disturbance combined by the friction force and cutting force of feed
system. u (k ) and wd (k ) are input signal and the noise of external disturbance respectively. A, B,
C, W, V are the matrixes of state equation.
Thus there is:
Bd 0
Ad Bd
Bd
1 0 0
0 , V 1 0 ,
, B ,C
,W
Α
0 1
0 1 0
0 0 1
0
0 1
1 e k bs TS
0
( u ) 2
,
,
R
A d B T / J , Bd
Kt Ts / J e
12
-1)/Be u
1 e e S e
K t (e
u is the quantization error of digital-to-analog conversion in the CNC system position
controller. (k ) is the expression of speed. m (k ) is the measured value of speed. d (k ) is
(k ) is the measured noise of speed.
disturbance. u (k ) is the measured noise of current,
Equation (3) is conversed into the discrete form of Kalman filter as follows:
A Force Monitoring Method Based on Built-in Sensors for Feed System … (Guangsheng Chen)
348
ISSN: 1693-6930
x (k )
x (k 1)
(k ) (I K C) A
(k 1)
obs
d (k )
d (k 1)
xm (k )
(I K obs C)B[u (k 1)] K obs
m (k )
(4)
K obs is the gain matrix of filter constant, it can be obtained by the following iterative
formula,
P(k | k - 1) = AP(k | k - 1)A T + WR ω W T
T
T -1
K obs (k) = P(k | k - 1)C [R v + CP(k | k - 1)C ]
P(k | k - 1) = [I - K obs (k)C]P(k | k - 1)
(5)
In the above equation, P(k-1|k-1) is the state estimation covariance of k-1 times
iteration. P(k|k-1) is estimation error of the next iteration based on matrix A and input noise (W,
Rw). At this time, Kobs (k) is the maximum estimation gain under present noise. Using new
state estimation matrix, covariance is updated to P(k|k). Generally, initial set value is a large
number, namely,
P(0 | 0)= 2 I, 1 .
The initial value of Kobs is set as 0. After several times of iteration, the value becomes
stable.
4. Results and Analysis
4.1. Experiment Environment
The force monitoring experiment of the feed system is conducted on the grinding wheel
truing and dressing system of a large CNC grinding machine, and the research is made on the
force of components on axis Y. The CNC system of this machine is NUM 3050. The simulation
signals of current i and rotating speed n are obtained from the external monitoring port of the
servo driver on axis Y through Advantech data acquisition card. Wherein, rotating speed signal
can be got by difference operation of the position signal in servo driver, but such signal has
higher noise. Position signal of axis Y can be acquired through collecting 1VPP signal of the
grating ruler by IK220 data acquisition card of Heidenhain Company, the experiment Setup are
shown as Figure 2.
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
TELKOMNIKA
ISSN: 1693-6930
349
(a) Servo feeding system Grinding for wheel dressing system of large CNC grinding machine
(b) Data collecting device
Figure 2. The experimental platform of force monitoring in feeding system
The relevant parameters are as follows: K t 1.1N / A , L=5mm, Ts=0.001s. The mean
square error of current, position, disturbance, rotation speed noise are taken as
R u =7.761 10-9 , R x =4.967 10-7 , R wd =0.01 , R ww =0.05
respectively, which can be obtained through the mean square error of the relevant
signal collected under the no feed condition. Here,
Je=0.00718 N/m2, Be=0.02213 Nm (rad/s)-1.
After calculation, the Kalman filter coefficients are obtained as follows:
1 7.96 10-7 0
0
1 0 0
A 0
1
-0.15 , B 0.153 , C
,
0 1 0
0
0
0
1
A Force Monitoring Method Based on Built-in Sensors for Feed System … (Guangsheng Chen)
350
ISSN: 1693-6930
K obs
3.34 10-3 5.49 10-7
5.53 10-2 0.310
,
2.80 10-2 -0.371
4.2. Load Detection of Feed System under Complex Condition
To monitor the external loading force under complex condition, the testing path of axis Y
is selected as shown in Figure 3 (a). External load is derived from the grinding force and internal
friction force of grinding wheel. Figure 3 (b) presents the current and feed speed signal. Figure 3
(c) is data processing result. In this figure, the thick dot dash line is the loading force obtained
by Kalman filter, while thin solid line 2 is got by equation (2). By comparison, the trend of the two
methods is almost consistent; but the former one is obviously smoother than the later one.
Besides, it has higher signal-to-noise ratio, and can be more easily applied in engineering.
(a) Displacement curve in Experimental path
(b) Current and rotating speed signal
(c) Comparison of experimental results
Figure 3. The experimental results of feed force monitoring under complex condition
4.3. Application Examples: Friction Force Testing of Feed System
Friction force can reflect the lubrication condition of slide guide in feed system, as well
as the installation error of slide guide and pre-tightened force state of movement components
etc. [8, 9] indirectly. Thus to evaluate the performance of the work condition of feed system, it is
required to detect the friction force of feed system.
When the machine is in no-load operation, Tc 0 , namely Te T f . In this condition, the
feed system of machine tool is in a reciprocating motion in different speed. By Kalman filter,
motor rotation speed and friction force can be obtained, and the Stribeck friction force curve of
feed system can be conveniently drawn, as shown in Figure 3 [10, 11].
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
TELKOMNIKA
ISSN: 1693-6930
351
Figure 4. Stribeck curves of friction force in feed system
5. Conclusions
This paper proposes a real-time method for using built-in sensor signals of CNC to
monitor the forces (including loading force and friction force) during the machining process of
CNC machine tool. This method can work without external sensor and be applied in production
site. Moreover, it can realize the optimal state observation of the force, motor rotation speed and
displacement of feed system using Kalman filter, and offer a way for determining filter
coefficients. Some experiments were carried out to verify the effectiveness of this method. The
results show that this method can perform the real-time and on-line monitoring to feed system
force state of CNC machine tool, and overcome the calculation noise in difference operation by
conventional methods. Moreover, the method can also provide manufacturers and users with
quantitative basis on the performance evaluation and fault diagnosis of feed system.
Acknowledgments
This project was supported by the Natural Science Foundation of Shanghai in China
(grant no. 13ZR1427500), the Key Projects in the National Science & Technology Pillar
Program of China (grant no. 2012BAF01B02), and the National Natural Science Foundation of
China (grant no. 51005158).
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
Cheng MY, Su KH, Wang SF. Contour error reduction for free-form contour following tasks of
biaxial motion control systems. Robotics and Computer-Integrated Manufacturing. 2009; 25(11):
323-333.
MN Khoshrodi, M Jannati, T Sutikno. A Review of Wind Speed Estimation for Wind Turbine
Systems Based on Kalman Filter Technique. International Journal of Electrical and Computer
Engineering (IJECE). 2016; 6 (4); 1406-1411.
Lima RM, Carvalho D, Vaccaro G, Scavarda LF. Industrial engineering and operations
management–special issue. International journal of industrial engineering and management
(IJIEM). 2013; 4(3): 103-108.
Zhou YQ, Mei XS, Jiang GD. Technology on Large Scale Numerical Control Machine Tool
Condition Monitoring Based on Built-in Sensors. Journal of mechanical engineering. 2009; 45(4):
125-131.
Q Zhang, S Xue. An improved multi-objective particle swarm optimization algorithm. Lecture Notes
in Computer Science. 2007; 4683(1): 372-381.
Yan X, Wu Q, Zhang C, Li W, Chen W, Luo W. An improved genetic algorithm and its application.
Indonesian Journal of Electrical Engineering and Computer Science. 2012; 10(5): 1081-1086.
Erkorkmaz K, Altintas Y. High speed CNC system design. Part II: modeling and identification of
feed drives. International Journal of Machine Tools & Manufacture. 2001; 41(10): 1487-1509.
IM Alsofyani, NRN Idris, T Sutikno, YA Alamri. An optimized Extended Kalman Filter for speed
sensorless direct troque control of an induction motor. 2012 IEEE International Conference on
Power and Energy (PECon). 2012: 319-324.
A Force Monitoring Method Based on Built-in Sensors for Feed System … (Guangsheng Chen)
352
[9]
ISSN: 1693-6930
Hamilton TW, Melbourne WG. Information content of a signle pass of Doppler data from a distant
spacecraft. JPL Space Programs Summary. 2003; 37-39(3): 18-23.
[10] IM Alsofyani, NRN Idris, YA Alamri, T Sutikno, A Wangsupphaphol, NM Nordin. Comparison of
Estimated Torques Using Low Pass Filter and Extended Kalman Filter for IM Drives at Low Speed.
International Journal of Power Electronics and Drive Systems. 2015; 6(1): 92-99.
[11] LI C, Cheng XG. A new approach for curve fitting based on the basic theory of recursive function.
Journal University of Shanghai for Science and Technology. 1998; 20(3): 226.
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
ISSN: 1693-6930, accredited A by DIKTI, Decree No: 58/DIKTI/Kep/2013
DOI: 10.12928/TELKOMNIKA.v14i3A.4429
345
A Force Monitoring Method Based on Built-in Sensors
for Feed System of Computer Numerical Control
Machine Tools
Guangsheng Chen*, Qingzhen Zheng
School of Mechanical Engineering, University of Shanghai for Science and Technology, Shanghai City,
200093, P. R. China
*Corresponding author, e-mail: cgs-168@163.com
Abstract
As force has important influence to the performance and machining quality of machine tool during
machining process, this paper proposed a method to monitor the load force of feed system using build-in
sensor of the machine tools. To reduce the influence of noise caused by external disturbance and
difference operation, a state observer based on Kalman filter that optimized values of force, velocity and
displacement of feed system is adopted, and gain coefficient of the observer is obtained by recursive
algorithm. To validate the effectiveness of the method, friction force testing experiments under no-load and
force testing experiments under complex cutting conditions are performed respectively, the results show
that data obtained by the method are effective, and this method significantly restrained the external noise
and noise caused by difference operation. The method suits to on-line and real-time force monitoring in
production without additional sensors.
Keywords: Kalman filter, feed system, force monitoring, built-in sensor
Copyright © 2016 Universitas Ahmad Dahlan. All rights reserved.
1. Introduction
Large CNC machine tools are key equipment in civil and national defense industry for
machining precision parts which are complex, large-scale, thin-walled and hard to machine.
High speed and high precision are the important development directions of CNC machines.
However, accuracy and stability problems arise as the increasing of the rotating speed of the
spindle and the feeding speed of servo system. High feed rate means the enlargement of the
system tracking error. To reduce the tracking error and improve the machining accuracy of
machine tools, researchers have taken some measures on the controller of machine tool, such
as improving the position and speed gains of the machine controller and using the feed forwards
of speed or acceleration etc. [1, 2]. However, the high-gain parameters affect the stability of the
control system. In addition, the stability problem of high-speed cutting, i.e. the cutting
instability(such as chatter) in the processing process in the case of large cut depth, high rotation
speed of the spindle, and high feeding speed condition, often plague the researchers and users.
On the other hand, when a large CNC machine tools work under unconventional conditions
such as high velocity, high acceleration, large load, large displacement, forces(including cutting
force, sliding friction ) of feed system are likely to cause vibration, impact, deformation, and
wear of components during machining ,thereby, processing system, spindle system and feed
system of CNC are significantly influenced. To meet the requirements including high precision,
high efficiency, and high performances of service as well as being intelligent under poor
machining condition, it is necessary to monitor load force on-line and real-time. The forces in
processing system can be obtained using tri-axial force sensor [3], however, as an external
sensor, which is often used in experimental studies and difficult to be applied in industrial
conditions because of inconvenience in installation. ZHOU Yuqinq [4] and RE Haber [5] et al
proposed that force conditions can be indirectly reflected by current signal of servo motor of the
machine, then requirements for performance evaluation and intelligent control of large CNC
machine tool can be achieved by monitoring and detecting the forces of feed system. However,
this method is only applied in some detection occasions due to the constraint of monitor
principle.
Received April 9, 2016; Revised July 24, 2016; Accepted August 6, 2016
346
ISSN: 1693-6930
This paper proposes a state observer based on Kalman filter. The observer can realize
the optimal estimation of feed system force state by collecting the built-in sensor signals of feed
system, such as motor current and motor encoder (or grating ruler) displacement etc.. Thereby,
the various states (no-load, online machining etc.) of CNC can be monitored conveniently.
2. The Monitoring Principle of Force in Feed System Based on Motor Current Signal
Typical feed system of CNC machine tool is comprised by amplifier, servo motor, ball
screw, screw nut, slide guide and worktable. Ball screw is driven by motors to make screw nut
move. Thereby, the rotating motion of the motors is transformed into linear motion. Worktable is
supported by sliding or rolling slide guide. And servo motor and ball-screw are directly
connected by couplings. To simplify the control model, the elasticity of the worktable is not
considered, and the mass and damping of all moving parts are equivalent to the equivalent
inertia and equivalent damping of motor shaft. The simplified electromechanical coupling control
system model of the feed system is showed in Figure 1 [6]. And the symbols in Figure 1 are
defined as the follows: u: the input signal of the controller, i: the current of the AC servo motor,
ym: the equivalent displacement of screw nut, l: screw lead, Kbs: the transferred coefficient of
the nut displacement corresponding to motor rotation angle, θ: the rotation angle of motor, ω :
the rotation velocity of motor, Kpp: the proportional gain of position loop, Kvp: the gain of
velocity loop, Kcp: the gain of current loop, Kt : motor torque constant, L and R: equivalent
inductance and equivalent resistance of motor respectively, s: Laplace transform operator, Je :
equivalent inertia, Bm: equivalent damping.
Figure 1. Simplified electromechanical control system of feed system
For the motor shaft, the equation is:
g
iK t = J e ω+ Be ω + Te
in the equation (1),
working state:
Te
(1)
is the equivalent torque of external load to the motor shaft, in
Te = T f + Tc
Tf is the friction in feed system, and Tc is the cutting force of the feed system in the feed
direction, so equation (1) equals to:
Te = iKt - J e ω- Be ω
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
(2)
TELKOMNIKA
ISSN: 1693-6930
347
If Kt, Je, Be, i, ω , ω were known, forces of the feed system (including external cutting
force and friction force) can be obtained by equation (2). In this equation, Kt can be obtained
from to motor manual, Jeand Be can be acquired by identification or calculation. However, ω and
ω
in the equation can not be obtained directly. To obtain the two values, usually, we should
firstly get the motor displacement sensor (encoder or grating ruler) signal y. By the difference
operation of y, ω is obtained. Then through the differential operation of ω , ω is acquired.
However, due to the error amplification of difference operation, especially in the real-time
monitoring system with sampling time on millisecond level, the signal-to-noise ratio of ω and ω
is sharply weakened. Especially after 2 times of difference operation, signal ω is almost
covered by calculation noise. Thus the detected force signal is provided with very low signal-tonoise ratio which even can be meaningless for research. To accurately monitoring the external
forces of feed system, a state observer based on Kalman filter is designed for conducting
optimal value observation to motor rotation speed and external torque.
3. Design of Kalman Filter
Kalman filter algorithm is proposed by American scholar Rudolf. Kalman. It is an optimal
forecasting estimation method based on minimum variance and can provide solutions to the
random noise disturbance in measured signal. Equation (1) can be written as discrete state
equation [7] which includes external disturbances, measurement and input noise.
x ( k 1)
x(k )
( k 1) A ( k ) B[u ( k )] W u ( k )
wd ( k )
d ( k )
d ( k 1)
x(k )
m ( k )
( k ) V
(k )
C
d m (k )
d ( k )
(3)
d (k ) is the force disturbance combined by the friction force and cutting force of feed
system. u (k ) and wd (k ) are input signal and the noise of external disturbance respectively. A, B,
C, W, V are the matrixes of state equation.
Thus there is:
Bd 0
Ad Bd
Bd
1 0 0
0 , V 1 0 ,
, B ,C
,W
Α
0 1
0 1 0
0 0 1
0
0 1
1 e k bs TS
0
( u ) 2
,
,
R
A d B T / J , Bd
Kt Ts / J e
12
-1)/Be u
1 e e S e
K t (e
u is the quantization error of digital-to-analog conversion in the CNC system position
controller. (k ) is the expression of speed. m (k ) is the measured value of speed. d (k ) is
(k ) is the measured noise of speed.
disturbance. u (k ) is the measured noise of current,
Equation (3) is conversed into the discrete form of Kalman filter as follows:
A Force Monitoring Method Based on Built-in Sensors for Feed System … (Guangsheng Chen)
348
ISSN: 1693-6930
x (k )
x (k 1)
(k ) (I K C) A
(k 1)
obs
d (k )
d (k 1)
xm (k )
(I K obs C)B[u (k 1)] K obs
m (k )
(4)
K obs is the gain matrix of filter constant, it can be obtained by the following iterative
formula,
P(k | k - 1) = AP(k | k - 1)A T + WR ω W T
T
T -1
K obs (k) = P(k | k - 1)C [R v + CP(k | k - 1)C ]
P(k | k - 1) = [I - K obs (k)C]P(k | k - 1)
(5)
In the above equation, P(k-1|k-1) is the state estimation covariance of k-1 times
iteration. P(k|k-1) is estimation error of the next iteration based on matrix A and input noise (W,
Rw). At this time, Kobs (k) is the maximum estimation gain under present noise. Using new
state estimation matrix, covariance is updated to P(k|k). Generally, initial set value is a large
number, namely,
P(0 | 0)= 2 I, 1 .
The initial value of Kobs is set as 0. After several times of iteration, the value becomes
stable.
4. Results and Analysis
4.1. Experiment Environment
The force monitoring experiment of the feed system is conducted on the grinding wheel
truing and dressing system of a large CNC grinding machine, and the research is made on the
force of components on axis Y. The CNC system of this machine is NUM 3050. The simulation
signals of current i and rotating speed n are obtained from the external monitoring port of the
servo driver on axis Y through Advantech data acquisition card. Wherein, rotating speed signal
can be got by difference operation of the position signal in servo driver, but such signal has
higher noise. Position signal of axis Y can be acquired through collecting 1VPP signal of the
grating ruler by IK220 data acquisition card of Heidenhain Company, the experiment Setup are
shown as Figure 2.
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
TELKOMNIKA
ISSN: 1693-6930
349
(a) Servo feeding system Grinding for wheel dressing system of large CNC grinding machine
(b) Data collecting device
Figure 2. The experimental platform of force monitoring in feeding system
The relevant parameters are as follows: K t 1.1N / A , L=5mm, Ts=0.001s. The mean
square error of current, position, disturbance, rotation speed noise are taken as
R u =7.761 10-9 , R x =4.967 10-7 , R wd =0.01 , R ww =0.05
respectively, which can be obtained through the mean square error of the relevant
signal collected under the no feed condition. Here,
Je=0.00718 N/m2, Be=0.02213 Nm (rad/s)-1.
After calculation, the Kalman filter coefficients are obtained as follows:
1 7.96 10-7 0
0
1 0 0
A 0
1
-0.15 , B 0.153 , C
,
0 1 0
0
0
0
1
A Force Monitoring Method Based on Built-in Sensors for Feed System … (Guangsheng Chen)
350
ISSN: 1693-6930
K obs
3.34 10-3 5.49 10-7
5.53 10-2 0.310
,
2.80 10-2 -0.371
4.2. Load Detection of Feed System under Complex Condition
To monitor the external loading force under complex condition, the testing path of axis Y
is selected as shown in Figure 3 (a). External load is derived from the grinding force and internal
friction force of grinding wheel. Figure 3 (b) presents the current and feed speed signal. Figure 3
(c) is data processing result. In this figure, the thick dot dash line is the loading force obtained
by Kalman filter, while thin solid line 2 is got by equation (2). By comparison, the trend of the two
methods is almost consistent; but the former one is obviously smoother than the later one.
Besides, it has higher signal-to-noise ratio, and can be more easily applied in engineering.
(a) Displacement curve in Experimental path
(b) Current and rotating speed signal
(c) Comparison of experimental results
Figure 3. The experimental results of feed force monitoring under complex condition
4.3. Application Examples: Friction Force Testing of Feed System
Friction force can reflect the lubrication condition of slide guide in feed system, as well
as the installation error of slide guide and pre-tightened force state of movement components
etc. [8, 9] indirectly. Thus to evaluate the performance of the work condition of feed system, it is
required to detect the friction force of feed system.
When the machine is in no-load operation, Tc 0 , namely Te T f . In this condition, the
feed system of machine tool is in a reciprocating motion in different speed. By Kalman filter,
motor rotation speed and friction force can be obtained, and the Stribeck friction force curve of
feed system can be conveniently drawn, as shown in Figure 3 [10, 11].
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352
TELKOMNIKA
ISSN: 1693-6930
351
Figure 4. Stribeck curves of friction force in feed system
5. Conclusions
This paper proposes a real-time method for using built-in sensor signals of CNC to
monitor the forces (including loading force and friction force) during the machining process of
CNC machine tool. This method can work without external sensor and be applied in production
site. Moreover, it can realize the optimal state observation of the force, motor rotation speed and
displacement of feed system using Kalman filter, and offer a way for determining filter
coefficients. Some experiments were carried out to verify the effectiveness of this method. The
results show that this method can perform the real-time and on-line monitoring to feed system
force state of CNC machine tool, and overcome the calculation noise in difference operation by
conventional methods. Moreover, the method can also provide manufacturers and users with
quantitative basis on the performance evaluation and fault diagnosis of feed system.
Acknowledgments
This project was supported by the Natural Science Foundation of Shanghai in China
(grant no. 13ZR1427500), the Key Projects in the National Science & Technology Pillar
Program of China (grant no. 2012BAF01B02), and the National Natural Science Foundation of
China (grant no. 51005158).
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
Cheng MY, Su KH, Wang SF. Contour error reduction for free-form contour following tasks of
biaxial motion control systems. Robotics and Computer-Integrated Manufacturing. 2009; 25(11):
323-333.
MN Khoshrodi, M Jannati, T Sutikno. A Review of Wind Speed Estimation for Wind Turbine
Systems Based on Kalman Filter Technique. International Journal of Electrical and Computer
Engineering (IJECE). 2016; 6 (4); 1406-1411.
Lima RM, Carvalho D, Vaccaro G, Scavarda LF. Industrial engineering and operations
management–special issue. International journal of industrial engineering and management
(IJIEM). 2013; 4(3): 103-108.
Zhou YQ, Mei XS, Jiang GD. Technology on Large Scale Numerical Control Machine Tool
Condition Monitoring Based on Built-in Sensors. Journal of mechanical engineering. 2009; 45(4):
125-131.
Q Zhang, S Xue. An improved multi-objective particle swarm optimization algorithm. Lecture Notes
in Computer Science. 2007; 4683(1): 372-381.
Yan X, Wu Q, Zhang C, Li W, Chen W, Luo W. An improved genetic algorithm and its application.
Indonesian Journal of Electrical Engineering and Computer Science. 2012; 10(5): 1081-1086.
Erkorkmaz K, Altintas Y. High speed CNC system design. Part II: modeling and identification of
feed drives. International Journal of Machine Tools & Manufacture. 2001; 41(10): 1487-1509.
IM Alsofyani, NRN Idris, T Sutikno, YA Alamri. An optimized Extended Kalman Filter for speed
sensorless direct troque control of an induction motor. 2012 IEEE International Conference on
Power and Energy (PECon). 2012: 319-324.
A Force Monitoring Method Based on Built-in Sensors for Feed System … (Guangsheng Chen)
352
[9]
ISSN: 1693-6930
Hamilton TW, Melbourne WG. Information content of a signle pass of Doppler data from a distant
spacecraft. JPL Space Programs Summary. 2003; 37-39(3): 18-23.
[10] IM Alsofyani, NRN Idris, YA Alamri, T Sutikno, A Wangsupphaphol, NM Nordin. Comparison of
Estimated Torques Using Low Pass Filter and Extended Kalman Filter for IM Drives at Low Speed.
International Journal of Power Electronics and Drive Systems. 2015; 6(1): 92-99.
[11] LI C, Cheng XG. A new approach for curve fitting based on the basic theory of recursive function.
Journal University of Shanghai for Science and Technology. 1998; 20(3): 226.
TELKOMNIKA Vol. 14, No. 3A, September 2016 : 345 – 352