Application of Fuzzy-PI Controller with

Feedforward Control in Direct Current Motor

Servo System
ZHOU Runjing, DU Yu, Yuan Weiting
Departmnent of Automation
College of Sciences and Technology
Inner Mongolia University
Hohhot, PRC
E-mail:auzhouj@ l63.com
Abstract-This paper provides a design of a kind of oripgnal paper has many advantages of feedforward control, fuzzy
fuzzy-PI controller that has compensating feedforward and an control and PI control of classic control theory, which not
introduction to the relative principle of arithmetic design. only satisfies dynamic and static demand of system, but also
Arithmetic realizing with VHDL is embedded into FPGA to restrains influence of the nonlinear factors. It has the
implement position loop control, velocity loop control and advantages of better robustness and not dependent on
current loop control. Furthermore we simulate the system
with Matlab and Simulink, finally present the actual tracking accurate-mathematic model, so it improves greatly dynamic
result to prove that it is a kind of high-performance of direct and static performance of system. An effective method for
motor servo controller. the servo system control with nonlinear and uncertain
factors is presented.
I . INTRODUCTION
HI. CONTROL PRINCIPLE OF SYSTEM
Direct current brush motor often is used to be the driver
of servo system for its characteristic of small volume and The system includes three loops of position feedback,
strong driving capacity. The controller designed in this velocity feedback and current feedback. See Figl:
r - ~~- -- - - - - -.._.
servo motor simple model
I I

Fig.l . System control principle diagram

In system the function of current loop is to restrain the In this system, feedforward fuzzy-PI control arithmetic is
interference of internal armature current and limit maximum effectively the respond speed of input signal and eliminate
current to protect motor. Velocity loop mainly restrain partly integral hysteresis effect, thus we can rapidly
velocity fluctuation to improve the ability against load decrease the error and increase the available bandwidth of
disturbing. Position loop is the primary control loop to system. In feedback control, fuzzy control is used for
realize position tracking. Three loops working together improving the dynamic respond speed at larger error range
ensure great dynamic and static characteristic of system. and PI control is adopted to improve static characteristic at
smaller error range.
I.ARITHMETIC DESIGN AND SYSTEM SIMULATION A.Feedforward control principle
applied. The purpose of feedforward control is to improve According to invariability principle, we can get:

0-7803-9422-4/05/$20.00 C2005 IEEE

1262
 r d r 1 ar "IF A and B THEN C" format to establish fuzzy rules,
F (t) =
K KC d2 t Ke K dt such as:
if e is PB and e' is PS, then u is PB
Discrete difference equation is: Namely if position error is PB and error variety rate is
PS,then electromotor turn fleetly towards positive
ul (k) = K1Ar(k) + K2Ar (k) position.By analogy with above inferences, other rules can
In the above equation: be translated to a fuzzy relation. Thus we get the control
Ar(k) = r(k) - r(k -1) rules of system that shows in the Table I.
Ar (k) =Ar(k) Ar(k 1)
- -

K1 = 1 /(KeKc AT) PB PS Z NS NB
K2 = r /(KeKcAT );
2 PB NB NM NM NS PM
PS NB PM PM PS PB
Z NB NS Z PS PB
KI, K2 is decided by character of controlled object, in this NS NB Z PS PS PB
system we specify the following two parameters according NB NM PS PM PM PB
to simulation results:
K,= 4, K2= 0. 5 Table I. Fuzzy control rules table
B. Fuzzy-PI controlprinciple In this system centroid method is utilized that is most
When servo motor startup main load is friction. Because popular and important to all defuzzification methods, and it
its mathematic model is difficult to decide, we use can represent with discrete formula:
characteristic of fuzzy control that don't need accurate u
*
=
4(ui)ui
model of controlled object, then we can realize fast startup pP(Ud)
and track.
We take displacement error (e) and error variety rate (e') as C. PI control principle
fuzzy inputs and PWM control signal as output. Fuzzy
controller utilize the different effect of factors Ke, Kec and Because fuzzy control can't eliminate the static error of
Ku of above three variables to system performance and system, we introduce PI control.
adjust three factors according to the different control Proportion control reflects error signal proportionately,
condition. The basic principle is that when error is big when error happens controller promptly generates control
weight of error also is big to eliminate the displacement signal to decrease the error.
error quickly; when error is small error variety rate is big to Integral control mainly eliminates the static error to
decrease the overshoot and make system steady quickly; Ku advance control precision.
can be regard as generally magnified multiple of system. In this system we adopt integral separate PI arithmetic to
After setting the reasonable factors we realize the avoid integral saturation. So when error is more than 0.5V
optimization of system. controller only calculate the proportion factor, otherwise
We use five language variables to define e, ec and u : NB, when error is less than 0.5V controller introduce integral
NS, Z, PS and PB. The basic discussion range are all [-5, 5]. factor.
Membership is showed by Fig 2.
D. The model ofcontrolled object
The parameters of direct current motor are: rotate speed
without load is 4100 rad/m, decelerated ratio is 1/160 and
rating voltage is 56V. Omitting inductance of armature and
viscosity damp factor we obtain the concise transfer
function of direct current servo motor that voltage ua (t) of
armature is input variable and rotate speed of motor is
output variable:

Fig.2. Membership function of input and output signal

Using fuzzy condition and fuzzy relation

1263
 Zero-Order
Hold

Fig.3. System diagram

CDIn 1 Derivrative2 Derivativel Oain7

Fig. 4. Feedforward control simulation diagram

1I Ke
H (s) =
Tms + 1
In this fornula, back electromotive force factor of motor
160 x 60 x 56
K =1 time constant Tm=lOms, Feedback
4100x 360 FL, -
duodt
factor KF = 1sV / 131.4 that we can regard it as angle Inl St Dt sFiUonre
sensor in the practice, but the transfer function represent
the relation between voltage and angle, so we add a
integral element to realize the transfer relation between Fig.6. Fuzzy control diagram
voltage and angle.
E. System simulation
According to above discussions we get below
AN simulation diagram with Matlab and Simulink. See Fig7
and Fig8.
Scope2 Even there are some model dismatch errors due to all
kind of reasons the influence is also smaller because this
WInl )
Gain3
Gain3
+
error can be resolved by fuzzy-PI controller. The system
X _1
u
i not only has the capacity of quick respond and excellent
anti-jamming, so it is very fit for direct current servo
Integrator GainG system processing nonlinear and uncertain characteristic.
Fig.5. PI control diagramn

1264
 Fig.7. Sine wave tracking without feedforward control

I~~~~~~. ... . . ........ . . . .. ... .. .. .... . . ...h\

cig.8a..
F stg
i ine ..tracking feedforward control
.

:Track gna

Fig.8. Sine wave tracking with feedforward control

IV. SOFTWARE DESIGN
The waveforms are measured to make it a condition that
According to simulation result with MATLAB we get load moment is 50N-m and the motor supply voltage is 56V.
several parameters. VHDL (Very High Speed Hardware For sine wave, square wave and optional wave, this system
Description Language) is used to implement arithmetic has better tracking characteristic, such as low overshoot and
above and program is embedded into FPGA (Field short adjustment time. The overshoot of lHz sine wave
Programmable Gate-Array) as real controller. IDE is MAX with lOV peak to peak has scarcely shoot and its phase
+plus II of Altera development systems. delay is 12'. But the overshoot of 2Hz sine wave with 1OV
peak to peak is about 2% of peak and its phase delay is 40'.
V CONLUSION
We can see with frequency analyzer, that the 3dB
bandwidth of system is 3Hz when instruction signal is
IOV sine wave. The following five figures are some
The adopted Controller is FLEXIOKlOOLC-3 of Altera's measuring results that we obtained from experiments.
FLEX 1OK device family. Test instrument is the data
acquisition system of Odyssey Corporation.

Fig.9. Tracking of 1Hz sine wave with lOV peak to peak

1265
Fig.1O. Tracking of 2Hz sine wave with 1OV peak to peak

Fig.1. Tracking of 0.5Hz square wave with 1OV peak to peak

Fig.12. Tracking of optional wave

1266
 Fig.13. Software design flow
ACKNOWLEDGMENT REFERENCES

This paper proposes a kind of control strategy that [1] Wang Lixin. A course in fuzzy systems & control
integrate individual advantages of fuzzy control, PI [M].Tsinghua University Press,2003,5, pp. 196-204
control and feedforward control, so we get a compound [2] Timothy J.Ross. Fuzzy logic with engineering
controller that can realize effective control for nonlinear applications [M]. Publishing House of Electronics
object, specially use FPGA in this system to greatly Industry, 2001,11:195-208, pp. 311F-320
improve the high-speed respond capacity. In conclusion it [3] Pan song, Huang jiye, EDA technique tutorial [M].
provides a kind of effective method to Science Press, 2002, 7:160-171
develop high precision and high performance of direct [4] Song zhiyong, Digital control and electronic control
servo control system. technique[M], China Machine press,2002,1:156&176

1267