FOURTH INTERNATIONAL SEMINAR AND WORKSHOP

In association with the

Engineering Design in Integrated Product Development

MECHATRONIC APPROACH TO THE MACHINE TOOL DESIGN

Z. GOSIEWSKI, M. SZAFARCZYK
Warsaw University of Technology
Department of Manufacture Engineering
[email protected], [email protected]

Keywords: Machine tools, mechatronic design, robust control, drives, vibrations

Abstract: In the paper we consider the possibilities of complex approach to the design of ma-
chine tools in the light of the newest means which the designer can use on the earlier stages of the
design procedure. Same of components is delivered with the controllers in the form of the micro-
processors. In that case it is useful to change nominal software to adjust the control law to our
purposes. So more we can add additional control loops to improve dynamics of the machine

1. INTRODUCTION causes that all elements of tool machine form one

dynamical system influenced by control systems.
Machine tools seem to be a very good example In modern machine tools each drive is a servo-
for discussing the influence of modern electronics, mechanism with control loop containing: sensors,
sensors and sophisticated drives on “mechanical” digital controllers, and actuators. To improve ma-
design. As a matter of fact mechatronic, as a sym- chine precision we are also forced to consider the
biosis of mechanics, electronics, informatics..., was possibilities of the introduction of vibration and
born in the field of machine tools. It happened about movement control system in the form of feedforward
50 years ago when computers were introduced to the or/and feedback systems [2]. It means that dynamics
description of spatial shapes and to the automatic of mechanical structure is strongly coupled with
control of machine tools. CAD systems were in- dynamics of local or global control systems particu-
vented and introduced to industry because of the larly in the range of higher frequencies.
numerical control, NC, of machine tools. Different criteria are taking into account during
Looking from another side, numerical control designing of machine tools. For example the geo-
revolutionized the machine tools themselves. NC metric accuracy of the machning elements is one of
machine tools are quite different than no-NC ma- the important features of modern tools. The accuracy
chine tools used to be. Traditional machine tools depends on the compliance of machine and on the
were designed by their producers in practically every cutting forces. Cutting forces can be reduced by the
detail as a new, original construction, made from introduction of High Speed Cutting process. There is
new subassemblies. Modern NC machine tools are no simple way to reduce compliance. Of course, we
made of modules which in many cases are produced can rise the rigidity of the machine by increase of its
by specialized enterprises. Even many elements used mass and dimensions but this is in opposition to the
in the design of machine tools (rolling guideways, economy and need for high acceleration. Typical
ball screws...) are delivered by specialized producers way to decrease the compliance is the structural or
and selected from catalogues. parametric modification of the machine structure.
Design stage time of machine tools can be re- Such passive approach may be sufficient in the case
duced by using of mechatronic approach to their of slow machne tools were the static stiffness is the
conceptual analysis and preliminary synthesis. Dur- most important.
ing synthesis and analysis we should consider the In the case of faster machines beside static stiff-
complete machine taking into account mechanical ness the big role plays the dynamic stiffness of the
structure, drivers, control systems and cutting proc- system: machine-tool-machining element. Presently,
ess. The development of faster and faster machine we are able to influence machine stiffness in the
tools, to reduce machining time and assure the re- broad range of frequency by the active means. Mov-
quired precision, needs stiff but light mechanical ing parts cause that static and dynamic stiffness
structures and drives with wide bandwidth [1]. It changes during machine operation. Therefore, to
shape the dynamic characteristics the adaptive con-
trol methods are often used. Lately, more advanced
166 EDIProD’2004

control methods are applied to the mechanical sys- more flexible and easier for implementation of “intel-
tems. They are called robust methods since such ligence”. At the same time their stiffness may be
system is robust to the parameter changes and to different than in the case of pure mechanical connec-
external excitations. Robust controller has constant tions.
parameters and assures sufficient stability and per- An important possibility of “mechatronic ma-
formance of the closed-loop system in the broad chine tools” is reconfiguration. Modules bought
range of parameters of the operation points. from the catalogued items and used in modern ma-
In the paper we consider the possibilities of chine tools must have unified (and frequently stan-
complex approach to the design of machine tools in dardized) interfaces. It makes possible changing
the light of the newest means which the designer can configuration of the machine tool after some time of
use on the earlier stages of the design procedure. its use. The reconfiguration may change the machin-
Same of components is delivered with the control- ing parameters and allow the user of the machine
lers in the form of the microprocessors. In that case tool to be much more flexible and respond to the
it is useful to change nominal software to adjust the changing demands of the market quicker and with
control law to our purposes. So more we can add smaller expenses.
additional control loops to improve dynamics of the But the best way to reconfigure system is the
machine. Such problems will be considered in our introduction of proper software to the controllers. It
paper. allows the low cost modernization of the machine
tools. If we add the control systems of structure
2. CONFIGURATION OF MACHINE parameters we can design the smart machine tools.
TOOLS
3. MECHANICAL STRUCTURE
A designer of a new machine tool has much more
MODELING
freedom than a designer of a new car. In the car
design there are just a few configurations of the Classic motion equations of flexible or mass-
main modules and a designer has a limited choice at lumped structure with a finite number of degrees of
the stage of a conceptual design. In the field of ma- freedom are as follows:
chine tools there are many possible configurations of
modules which can fulfill: the required parameters Mx
&& + Cx& + kx = f (1)
of workpieces to be machined, NC movements and
additional functions. where x and f are the vectors of generalized dis-
The freedom of choice requires responsibility of placements (translations and displacements) and
a machine tool designer. It is very important for the forces (point forces and torques) and M, K and C are
final result what configuration of the machine tool has respectively the mass, stiffness, and damping matri-
been chosen at the conceptual stage of design. The ces, they are symmetric and semi positive definite.
know-how of the design team and a heuristic ap- M and K arise from the discretization of the struc-
proach may be not enough. A certain method of multi ture, usually with finite elements. A lumped mass
body simulation, BMS, is propose as a tool for the system has diagonal mass matrix while the finite
configuration choice based of estimation of the work- element method usually leads to non-diagonal mass
ing stiffness of different mechanical structures [3]. matrix. In the complex machine tools the lamped
In modern manufacturing monitoring is more mass system also have non-diagonal mass matrix
and more popular, frequently as a part of automatic [3].
supervision [15]. Automatic supervision attempts to The damping matrix C represents the various
eliminate the influence of disturbances, to guard dissipation mechanisms in the structure, which are
machine tool or/and workpiece and to optimize pro- usually poorly known. Therefore it is customary to
duction. The term monitoring was adopted from assume hypothesis about Rayleigh damping:
medicine where it is as a kind of automatic, simpli- C = α M + β K , where coefficients α, β are selected
fied diagnostics. Diagnostics, which is directed only to fit the structure under consideration.
at a chosen features of monitored object, but is ac- Let us introduce modal transformation:
tive all the time. In hospitals the special sensors are x = Φz , where z is the vector of modal coordinates.
mounted on seriously ill patients and the signals The transformation leads to decoupled modal equa-
from the sensors are continuously and automatically tions:
analyzed from the point of view of selected features
z + 2ξΩz& + Ω 2 = μ -1Φ T f
&& (2)
(e.g. pulse, blood pressure... ). When the value of
selected feature is identified as critical the hospital where: Ω = diag (ωi ) is the matrix of natural fre-
personnel is automatically called to intervene. quencies, μ = diag ( μi ) is the matrix of modal
Mechatronic modules with their own drive and 1 α
sensor units are connected by information signals masses, and ξ = diag (ξi = [ + βωi ]) is the matrix
with other mechatronic units and with the control 2 ωi
system of the machine tool. Such structure is much of model damping.
 EDIProD’2004 167

The transfer function between the force f as an As we show later the bandwidth can be
input and z as an output is in the following form: wider in the case of proper choice of control law. A
robust control system seems to be a perfect choice in
G (ω ) = ⎡⎣ −ω 2 M + jωC + K ⎤⎦ = ∑
−1 n φφ i i
T
the case of loaded machines. To design the robust
i =1 μ (ω − ω + 2 jξ ωω )
i i
2 2
i i
controller we should consider all interactions of the
(3) control plant with surroundings and take into acount
Typical Bode characteristics of the transfer function all spectrum of nonlinear plant parameters (which
(3) are shown in the Fig.1. are conected for example with operation points)
when we are designing of linear model of the control
plant [4].
The interactions and changes of parameters
are represented by weigth functions in th form of
transfer functions as it is shown in the Fig. 2. The
weight functions are chosen in such a way to create
the systems parameters which are able to fulfill the
assumptions. So, the weight disturbances Wo(s) i
Wi(s) describe the contents of the disturbances di i do
or they may be used to model disturbance power
spectrum which depends on the nature of signals
involved in the practical systems. The weight Wn(s)
describes the frequency model of the sensor noise.
From the opposite side, the weight We(s) reflects the
requirements put on the shape of the closed-loop
Fig.1. Typical frequency characteristics of a me- transfer functions, i.e. on the shape of the output
chanical structure. sensitivity function. Similarly, the weight Wu(s)
describes some restrictions put on the control or
For control purposes we usually reduce the model to actuator signals. The last weight Wr(s) is an optional
m lowest modes: element used to achieve desired shape of command
φ (k ) 2
or to represent a nonunity feedback system in an
G (ω ) = ∑
m
+R (4)
i

i =0 μ (ω − ω ) 2 2 equivalent unity feedback form [5,6].

i i
In summary, the main step in controller design
and residual modes are: procedure is to choose the appropriate weight func-
R = ∑ φφ
T
i i
n
(5) tions and to implement them in practical applica-
m +1 μω i i tions.
Residual modes in the form (5) cause that transfer
function (4) is non-proper function. [2]. It leads to
big differences between zeros of transfer functions
(3) and (4).
For ω=0 we obtain from (3) the modal
expansion of the static flexibility matrix:
n φφ T
G (0) = K −1 = ∑ i i
= Φ ( Φ T K Φ ) −1 Φ T (6)
i =1 μiωi

One of the aims of vibration control system of ma- Fig.2. Functional diagram of system for robust
chine tools is to make the static flexible matrix ele- analysis purposes.
ments as small as possible.
The way of choice of the weights is shown in Fig. 3
4. DRIVES and is described by equations (7). The control error
weight We(s) is associated with the sensitivity
Electric and hydraulic drives are used in machine function S(s). The sensitivity function |S| should
tools. Usully we assume that drives work perfectly. have the low value for the low frequencies. In this
It means that they have amplifications equal one (0 range the control signal and disturbance signal play
[dB]) in the sufficient bandwith.. It is true in the case a meaningful role. So, the sensitivity function must
of slowly working machine tools. In this case we can fulfill the requirement |We(s)⋅S(s)| ≤ 1, where the
omit the dynamics of the drives. We should take into weight We(s)⋅is described by equation (7). The form
account the dynamcs of servo-drives when the ma- of the weight is realized in systems with steady-state
chines work with acceleration reaching 3[g]. The error. In considered systems we assume the error
bandwidth of servo-drives depends on many causes: lower than ε. This is ensured when |S(0)| ≤ ε. So, for
loads, operation point, control law, external distur- |We(0)|≥1 the norm is ||We(s)S(s)||∞ ≤ 1. The similar
bances, and so on.
168 EDIProD’2004

analyses can be used for other weights and functions troller were optimised in Matlab by using of NCD
of the system. Blockset Toolbox [9]. As a result we obtain the
following PID controller parameters: Kp=1 – pro-
k
⎛ ωbc ⎞
k portional gain; Ki=0.25 – gain of integration part;
⎛ 1 ⎞
⎜ k M s + ωb ⎟ ⎜ s+ k ⎟ Kd=0.01 – gain of differential part.
⎜ Mu ⎟ We assume in our case that the following
We ( s ) = ⎜ s ⎟ Wu ( s ) = (7)
⎜ s + ωb k ε ⎟ ⎜ k ε1 s + ωbc ⎟ parameters change during the electric drive opera-
⎜ ⎟ ⎜⎜ ⎟⎟ tion: k E - voltage constant of drive; Tm - mechani-
⎝ ⎠ ⎝ ⎠
where: k ≥ 1, ωb – the cut-off frequency for function cal time constant; Te - electrical time constant; in
S(jω), ε – the lower restriction of S(jω), Ms – the range ±10% of nominal value. For example,
the upper restriction of S(jω), ωbc – the cut-off electrical time constant can have values from the set
frequency for function R(jω), {0.252, 0.28, 0.308}. Transfer function of closed-
ε1 – the lower restriction of R(jω), Mu – the loop system with any nonnominal parameters will be
upper restriction of R(jω). denoted by GP(s).
Let us introduce relative error in the follow-
ing form: e( jω ) = GP ( jω ) − G ( jω ) . The choice of the
G ( jω )
weight is determined by condition that for all fre-
quencies the Bode diagram of weight should be
above the plot of relative error. To do this we chose
the weight transfer function in the
−3
form: ω u (s ) =
1 .5 ⋅ 10 ⋅ s
Fig.3. Limitations put on the weights functions. 1.5 ⋅ 10− 7 ⋅ s + 1
The next considered weight is connected with
The nominal model of the plant together with the scaling of the controller output signal u. For
weight functions forms an extended model of the simplicity we assume a scalar weight where its mag-
open-loop system. For such model we can obtain the nitude is as follows: ω I = 7 .
optimal controllers using of commercial software, In our case after calculations and simplifications
for example, Matlab-Simulink. we have obtained the controller described by the
following transfer function:
4.1. Electric drive
49.86 ⋅ s 2 + 2600.11 ⋅ s + 9096.58
Modern control theory has recently been widely K (s ) = .
applied to design the controllers for electric drives s 3 + 773.35 ⋅ s 2 + 4141.89 ⋅ s + 4.14
[7,8]. The reasons for this are following: 1) the ne-
cessity of meeting increasingly stringent require- The results of computer simulations of closed-loop
ments on the performance of drive control systems, system for both controllers are shown in the Fig. 4.
ROZRUCH SILNIKA
2) easy access to modern power semiconductor 5000

switching components and microprocessors with 4500

which very sophisticated control strategies can be 4000

implemented at reasonable cost, 3) modern control 3500

theory has been extended and modified so that it is

OBROTY [obr/min]

3000
practically applicable to microprocessor-based drive 2500
control by taking into account physical constraints 2000
such as input delay times and input and state vari- 1500
able constraints. 1000 PID
The object of consideration is DC motor D – 500
Hinf

101. At first, analytical model of the open-loop sys-

0
tem is derived. Next, the classical PID controller is 0 0.2 0.4 0.6 0.8 1
CZAS [s]
1.2 1.4 1.6 1.8 2

designed to obtain a reference system. After some

Fig. 4. Time response of closed-loop system with
theoretical consideration the robust controller was
designed. Performance of the systems with both PID and H ∞ to step change in the reference angu-
controllers is investigated by computer simulation lar speed (n=0-4400 [rpm]).
and in laboratory experiments.
By designing PID controller and optimising its We wanted to confirm the simulation results by
parameters we obtain the template which will be a experimental investigation of electrical drive D-101.
reference for later simulations and experiments with Therefore, the program of lab researches is similar to
robust control system. The parameters of PID con- the program of computer simulations. For both con-
 EDIProD’2004 169

trollers the start phase during which electric drive piston and servo-valve displacements ΔXV=ΔX =0 ,
reaches its operation speed was recorded on the lab and for three flow rates.
stand and transient processes are shown in Fig.5. 5600
ZMIANA nzad Z 4400 NA 5400 [obr/min]

From these plots we have calculated the perform-

ance parameters presented in Table 1. 5400

PRĘDKOŚĆ OBROTOWA SILNIKA W CZASIE ROZRUCHU 5200

4500

OBROTY [obr/min]
4000
5000
3500
4800
3000
OBROTY [obr/min]

2500 4600
nzad
2000 PID
4400 Hinf
1500
nzad=4400 [obr/min]
1000 4200
PID -0.5 0 0.5 1 1.5
Hinf
500
CZAS [s]

ZMIANA nzad Z 4400 NA 3000 [obr/min]

0
0 1 2 3 4 5 6 7 8 9 10 4600
CZAS [s]
4400 nzad
Fig. 5. Start up phase of the electric drive for both PID
Hinf
controllers recorded in the lab stand (reference 4200

angular speed equals 4400 [rpm]) 4000

OBROTY [obr/min]
3800

Tab. 1. Performance parameters of experimental 3600

control systems. 3400

3200
Parameters
3000

W [J] Tn [s] tr [s] 2800

-0.5 0 0.5 1 1.5
Controller Control en- Rise Control CZAS [s]

ergy time time Fig. 6. Recorded control system responses to step

changes in the reference angular speed:
PID 165.07 0.35 5.21 a) decreasing from 4400 to 3000 [rpm],
b) increasing from 4400 to 3000 [rpm].
H∞ 168.08 0.79 1.46
All characteristics of linear model are similar in
On the base of performance parameters col- given the frequency range. In the case of nonlinear
lected in the Table 1 we can notice that the system model we can see big changes of the slope of char-
with PID controller has faster response, but it takes acteristics. The biggest differences between models
longer time to reach the zero offset. It results from are in the vicinity of frequency 7.5 [Hz].
too small gain of integration part of controller. The For the flow rate Q3L the system bandwidths in
PID controller was designed for linearized model of the case of robust controllers are wider, than for PID
DC motor while the real motor has a little other controller. The increase of the bandwidths reaches
dynamic properties. We can see that the robust con- 30 %. System with optimal robust controller has the
trol system with H∞ controller is much more insensi- best cut-off parameter and it is 2,65 [Hz]. The results
tive to the plant uncertainty. The systems with both for gain parameters are similar to the preface case.
controllers consumed similar amount of electric
energy during start-up phase.
We also watched how drive reacts to step
changes of reference angular speed. Behaviour of
the closed-loop system is presented in Fig.6. There
is no big differences between systems with both
controllers. So more, there are zero steady-state
offsets and smaller overshoots than ones which ap-
peared during computer simulations.
4.2. Electro-hydraulic servo-drive
The frequency characteristics of the system for
nonlinear and linear model of servo-drive are
presented in the Figure 2 for the following parame-
Fig. 7. Amplitude-frequency characteristics of the
ters: supply pressures P10 0 = P20 0 = 21 [MPa], main-
linear and nonlinear model of servo-drive.
170 EDIProD’2004

The most interesting results presented on the

Figure 5 are for the flow rate Q4L. The system with
H2 controller has the widest bandwidth and there
consists 7,25 [Hz]. It means it is almost twice (100
[%]) wider than in the case of PID system (3,8
[Hz]). The characteristics has the resonant pick for
3,2 [Hz]. It reaches 3 [dB].
Step responses of the closed-loop system with Fig.9. Beam with piezoelektric actuator and sensor.
different parameters and controllers are also differ-
ent (Henzel, 2004). The signals are unsatisfactory The beam with actuator and sensor forms an
for robust controllers (H2 i Hinf). The system was open-loop system. The open-loop system was identi-
strongly disturbed and the step responses were more fied in the fequency domain. The amplitude-
oscillatory. Similar tests were carried out for differ- fequency characteristics is shown in Fig. 10. We can
notice the resonances for frequencies: 10, 60,
ent supply pressures. In this case system with H∞opt
180[Hz] and the anti-resonanses for frequencies: 1.3,
controllers works properly.
20 i 150[Hz]. We adjusted mathematical model (Fig.
11) by proper choice of the damping coefficients ξ .
The lead member of the transfer function is con-
nected with dynamics of the amplifier used to supply
the piezoelectric actuator.

Fig.8. The frequency characteristics.

In summary, the servo-drive with optimal robust

controllers is the best in practical applications. Such
system has the best dynamic and static parameters.

5. VIBRATION CONTROL SYSTEM

Fig.10.Recorded amplitude-frequency characteris-
The main problem in structure vibration control tics of the open-loop system.
is the availability of the actuators with sufficiently
high control force and wide bandwidth. For vibra-
tion control of rotating parts (rotors) we use usually s 2
s + 0.5s + 5.29
2
s + 14s + 400 2
s + 10s + 22500
the magnetic bearings [10] or linear piezoelectric 150s + 1 2
s + 3s + 100
2
s + 15s + 3600 2
s + 4s + 32400

actuators [11]. For non-rotating structure to vibration

control we use “structure borne” actuators as: reac- Fig.11. Transfer function of the open-loop system.
tion wheels, control moment gyros, proof-mass
actuators, piezo strips, etc. [2]. For the further con- The Bode characteristics of the transfer function
siderations about machine tools vibration control we from Fig. 11. are presented in Fig. 12. We used
take into account the piezoelectric strip as an actua- Matlab and SISO Design Tool [13] to consider the
tor and as a measurement unit [12]. influence of different controllers on the dynamics of
closed-loop system. The characteristics of closed-
5.1. Piezoelectric vibration loop system (Fig. 14) with inertial controller are
control system presented in Fig. 13. Comparing Fig. 12 and 13 we
can notice that the reduction of the highest peak of
Consider cantilever beam as a model of a lathe tool amplitude approach 30 [dB].
with piezoelectric actuator and sensor as in Fig. 9. The second-order filter in the feedback loop as-
sures stronger damping and bigger roll-off (–
40dB/dek.) than in the case of inertial controller. But
its drawback is connected with increasing the sensi-
tivity of system stability to the parameters of identi-
fied model. We can notice it by the comparision of
 EDIProD’2004 171

root locus plots for both controllers: inertial (see Fig

17) and filter (see Fig. 18).

Fig. 16. Bode charakteristicsof the closed-loop sys-

tem second-order filter.

Fig.12. Bode characteristics of the identified model

of open-loop system.

Fig. 17. Root locus for the system with inertial con-
troller.

Fig. 13. Bode characteristics of the system with

inertial controller.

s 2 2 2
s + 0.5s + 5.29 s + 14s + 400 s + 10s + 22500
150s + 1 2
s + 3s + 100
2
s + 15s + 3600 2
s + 4s + 32400

1
s +1

Fig.14. Closed-loop system with inertial controller.

For comparision the the second-order filter

was used as a controller (Fig.15). The Bode charac-
teristics of the closed-loop system with such control-
ler are given in Fig. 16.
s 2 2
s + 14s + 400 2
Fig. 18. Root lotus of the system with a second-order
s + 0.5s + 5.29 s + 10s + 22500
150s + 1 2
s + 3s + 100
2
s + 15s + 3600 2
s + 4s + 32400 filter.

2
1
s + 2s + 1
5.2. Experimental results
Fig.15. Model of system with second-order filter as
the controller. The inertial controller was verified in experimental
way by using of it in the vibration control system of
The closed-loop system with inertial controller beam from Fig. 9. Durig transient response the
is awans stable since branches of plots are in the left control system was lock-in and the results are seen
half-plane for all controller gains. System with sec- in Fig. 19. It was calculated that logarithmic decre-
ond-order filter becmes unstable for some controller ment of damping increased from δ=0.2 to δ=0.51.
gains. For uncorrect model the system designed as The amplitude-frequency characteristics is pre-
stable one can appear unstable. sented in Fig. 20. We see the high level of the damp-
ing in the whole range of frequency. Characteristics
is similar to respective characteristics of the simu-
lated closed-loop system from Fig. 13.
172 EDIProD’2004

4. In machine tools with smart structures we can also

deform the system to reach desired location of tools
and machinig piece. So we can resign from high
rigity of machine to reduce mass and energy costs.
5. New tools like electrospindles with fast rotating
tools (120000 rev/min) release forces acting during
the cutting process. If they are equipped in magnetic
bearings there is some possibilites to change the
location of the tool. It gives additional degrees of
freedom and make the machine tools more flexible
to the production requirements.

Fig. 19. Impuse response of experimental beam. The

References
[1]. Maj R., Modica F., Bianchi G.: Machine Tools
control system starts after 2.4 s.
Mechatronic Analysis. Proc. APE, Warsaw
2004..
[2]. Preumount A.: Vibration Control of Active
Structures.
[3]. Śniegulska D., Klasztorny M., Szafarczyk M.:
Assessment of Machine Tool Dynamic Proper-
ties at the Early Stage of Design.Proc. AnnAr-
bor.
[4]. Gosiewski Z., Henzel M.: Robust Control Meth-
ods as New Approach in Control Systems. Proc.
5th International Carpathian Control Conference,
Vol.1, pp. 403-408, May 2004.
[5]. Skogestad S., Postlethwaite I.: Multivariable
Feedback Control. Analysis and Design. Chich-
Fig.20. Amplitude-frequency characteristics of the ester, John Wiley & Sons Ltd. 1996.
closed-loop system with inertial control- [6]. Zhou K., Doyle J.C.: Essentials of Robust Con-
ler recorded on the lab stand. trol, Prentice Hall, Inc., 1998.
[7]. Dote Yasuhiko: Application of Modern Control
6. CONCLUSIONS Techniques to Motor Control. Proceedings of the
Machatronic approach to design gives new possi- IEEE, Vol.76, No. 4, April 1988, pp.438-454.
bilites in the improvement of economic and techni- [8]. Gosiewski Z., Hanc D.: Robust Control of an
cal properties of machine tools. In the paper we Aircraft Electric Drive. Proc. Fourth Interna-
considered influence of the local control loops on tional Seminar on RRDPAE-2000, Warsaw, Po-
the machine dynamics. We omitted dynamics of the land, December 2000.
main body of the machine tools. The dynamics of [9]. Nonlinear Control Design Blockset User’s
complete machines will be considered in the future Guide Version 5. MathWork 1999.
works. After this early-stage consideration we can [10]. Gosiewski Z., Falkowski K.: Wielofunkcyjne
come to the following conclusions. łożyska magnetyczne. Monografia nr 19 z serii
1. It is evident that in the similar way we can use Biblioteka Naukowa Instytutu Lotnictwa.
global control loops to improve the dynamics of the Wydawnictwa Instytutu Lotnictwa, 2003.
system: machine-tool-machining piece. In this case [11]. Alizadeh A., Ehmann C., Schonhoff U., Nord-
it is desired to measure the localization of the tool. mann R.: Vibration Control of Flexible Rotors
2. The model-based control systems are better than Using Piezo Actuators as Active Bearing. Proc.
artificial inteligence based control systems. So more 2nd International Symposium on Stability Con-
we can resign from adaptive methods since they are trol of Rotating Machinery. Gdańsk, Poland,
sensitive to adjusting of models to actual system. August 2003, pp. 485-496.
The robust controllers seems to be a good solution to [12]. Zientalski K.: Piezoelektryczny układ
control the vibrations and motions of machine tools. sterowania drganiami belki. Praca magisterska
3. New materials like piezoelectrics or alloys with wykonana pod kier. Z. Gosiewskiego, WAT
shape-memory can be used as sensors/actuators in Warszawa, 2004.
the control loops. Since they can be located in any [13]. Real-Time Interface to Simulink RT131.
part of the machine tool as a parameter-distributed User’s Guide dSPACE. Version 3.0, Germany.
elements we can design so called “smart” structures. [14]. Real-Time Workshop. User’s Guide. The
Smart structure is used to shape the dynamic charac- Mathworks, Version 3.0, Germany.
teristics of machine. [15]. Szafarczyk (Editor): Automatic Supervision
in Manufacturing. Springer-Verlag 1994.