Comparative Performance of Adaptive PIFF-DBF

and Convetional PID on Buck Converter

SABAR DAMANIK
Electrical Engineering Departement of Brawijaya University
Malang, Indonesia

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

 Abstract—in essence, great limitation. Switching motor is follow as: KT B
papers was to show the use of converter could be robust to S m ( s )  I A (s)  M  M (s)
PIFF-DFB control adaptive the variation of the sircuit JM JM
same as and parameters, the input and
load disturbances by (4)
feedforward as well as
nonlinear controls. S m .J M ( s )  K T I A ( s )  B M  M
feedback on buck converters
where an exodus testing will
The state sppace (5)
averaging technique for the
be compared directly with S m . J M  B M  M
reference model and modeling of converters  IA
conventional PID including the effects of KT
(Proportional Integral parasitics is prrovided in.
Derivative) controller. The The output of the converter (6)
PIFF-DFB control adaptive is regulated by controlling
is a controller variation that  S . J  B M  M  (7)
is expected to reduce errors
the on and off duration of Figure 1. DC Motor S .L A . m M  
and speed up the response on
the switch. Pulse width  KT 
conventional PID controller modulation (PWM) di a RA KT 1
technique is used to  i a (t )   m (t )   SU T (.tJ)  B M  M 
output, conventional PID
generating PWM signal. In
dt LA LA  R ALA m M   K T 
controllers are commonly
 KT 
used controller variation, and factual applications, (1)
reference model is desirable converters are powered by d m K T B  S m . J M  B M  M  (8)
 i a (t )  M  m (tS) .L A .  
forms of a system that has a non-idealvoltage sources dt JM JM  KT
three-order system. To display instead of the constant dc  
error values on conventional voltages and are loaded
PID controller and PIFF-  S .J  B M  M 
DFB control adaptive, then
with electrical applications. (2)  R A  m M   K T 
Consequently, they  KT 
ISE (Integral Square Error)
experience some variations where  M is the angular
and IAE (Integral Absolute  S .J  B M  M  (9)
Error) are created. The in the input and load. position,  M is the rotor S .L A . m M  
comparison result of this Therefore, the closed loop
speed, I A is the motor  KT 
response is obtained from the feedback control is
simulation in the Simulink necessary to retain the current, constant B M of  S . J  B M  M 
part of Mathlab.  R A  m M   K T 
output the output voltage at viscous friction, J M is the  KT 
a desired level regardless of inertia of the rotor, T L is
these variations. Various then the transfer function
Keywords—Propotional,
controllers have been load torque, R A is is :
integral feedforward,
derivative feedback, adaptive, designed and analyzed, haunted resistance, L A is
buck converter. such as propotional integral armature inductance, K T KT
feedforward and derrivative
return electromotive force M LA J M
feedback and conventional
(emf) constant or constant  2
I. INTRODUCTION PID controller. torque and UT applied U T S  S ( L A BM  R A J M )  ( B
Buck converter is II. DC MOTOR voltage to the motor. In DC (10)
motors, electromagnetic
b  KT
the simplest power This section describes torque (TE) and backemf
electronic circuit that the dynamics of a DC motor LA J M
(ub) is proportional to the
performs the step-down dc- with a variable equation motor current and motor
to-dc power conversion. status set. Mathematical a1  L A B M  R A J M
speed, each. Stator
With the rapid development model very important in
of microprocessor understanding any physical
magnetic flux remains is a 2  BM R A  K T 2
essentially constant at all TABEL 1
technology., small-sized system. The dynamic model levels of the anchor current,
and long service life battery for a DC motor is derived therefore the DC motor PARAMETERS PMDC MOTOR
in portable personal from both electrical circuits torque velocity curve is
Parameter Value
electronic product, buck and mechanical equations linear.The transfer function
motion. The equations 3X10-3 N.m.s/r
converters are used in low of a DC motor is important
voltage condition. describing the for designing the proper 162.73 m.H

However, forward voltage characteristics of a DC controller that will result in 7.72

of diode may reduce a high performance system 1.25 N.m
conversion efficiency in improvement. The image 2.36X10-2 kg.m2
this situation. below is a DC motor that is 21.0790 m.H/
Buck converter modeled for use in the 7.8667 kg.m.r/N/s
controlled at steady state by simulation process.
linierization control based on (1)-(2) :
technology based RA K 1
smallsignal model. Due to SI A ( s )   I A ( s)  T  M (s)  U T ( s)
the nonlinear of switching LA LA LA
converter, linearization (3)
control technology has a
 C ( s)  ki 
parameters
b  of KP, Ki, and
 the plant model

v  s 3   a1  bk d  s 2  a 2  bk p s  
  k p   k d s  2 Kd so that
E ( s)  s  s  can a 2  the reference
a1 s follow v '  bs
(11) model. The block diagram
for adaptive PID is found in
b

K n 2 fig.4 and the MIT rule used 
bs.s 3   a1  bk d  s 2  a 2(31)
 bk p
to determine Kp, Ki and Kd 
s 2  a1 s  a 2 s 2  2 n s  that
2
n change according to  s 3   a1  bk d  s 2   a 2  bk p
(12)

the change of plant
parameters is found on


bs. k p s  k i  k d s 2 .b.R ( s 
s
 
   s 2   i s  1 
 kp i d


 1
(21)-(27).
K 

3

  a1  bk d  s 2  a 2  bk p s 
  s  2 2
i s  s  1
 2 n 
 n 
Figure 2. Output plant (13)
without controller y p (32)

On the output graph DC k p
motor has T 0.112 seconds to determine the value of
and ts of 0.5 seconds. bs
Kp, Ki and Kd can be used
the following formula :
3 2

s   a1  bk d  s  a 2  bk p s  bk 
III. CONTROLLER DESIGN
K*  kpK

There are three control Figure 4. Model of

(14) dk p (33)
models presented in this adaptive PID 
paper namely conventional i dt
PID, adaptive PID and
 * 1 2
kpK J  bs
  i . .1 
reference model. Which
then compared between the
2
s   a1  bk d  s  a 2  b
3 2

three models. (15) (21)
2 a1 J
A. Conventional PID i   
n a2  dk i (34)
To achieve 
effective performance in dt
the output speed with (16) (22)
  y p ( s)  y m ( s ) b
speed controller speed, 1 1   i . .1 
proportional, integral and  i d 
n 2

a2
s   a1  bk d  s  a 2  b
3 2

derivative (PID) are (23)
common for DC motor

speed. The determination (17) 1
of PID constants is done y p dk d (35)
i 
manually by calculating kp  dt
directly without any  *K (24)
adjustment method. The bs 2
d J   d . .1 
determination of PID
constants depends on the
(18)
dt
 

s 3   a1  bk d  s 2  a 2  
kp
characteristics of DC ki  b, a1 , a 2  unknown
motors that change over i (25)
time. The block diagram J J  am1  a1  b.k d
closed loop for  . am 2  a 2  b.k p
conventional PID is found
(19)   
k d  k p d am3  b.k i
in fig.3 and the transfer
function is found on (13). (26)
sgn(b)  then  b * (  ); maka  b *
(20) dk p J J  y p
  p   p . .
dt k p  y p k p
dk p (36)
Adaptive PID (27) 
Adaptive PID is y p u u ' v  v 'u
dt
a control design The   bs
k p v v2   i . .1 
determination of PID 3 2
s  am1 s  am 2 s  am
constants depends on the
characteristics of DC (28)
Figure 3. Model of
motors that change over
conventional PID
time. With no knowledge of 
u  k p s  k i  k d s 2 .b  (29)
plant parameters, this
adaptive PID can adjust the u '  .bs
(30)
dk i A. Nominal Simulation
 (37)
Referensi Model 1) Testing without load Adaptive PID and
dt Conventional PID results
  i . .1  3 2
b y m ( sR 
) ( s ) bm

2

y p1 (ss )  bm2 s  bm3 can follow and go steady in
the reference specified and
s  am1 s  am 2 s  amR(3s ) s 3  am1 s 2  am 2 s  am3 resemble the reference
dk d (39) model that has been
 (38) obtained in the previous
dt • am1 = 4,38
chapter.
bs 2
  d . .1  3 2

• am2 = 247,3
R(s)  y p (s)  2) Tracking
s  am1 s  am 2 s  am
• a3m3 = 855
In this simulation
• bm1 = 24,72 by changing the reference
or input with a  M
Base on (36), (37), (38) • bm2= 211,7 Figure 8. Conventional PID
change that increased 150
represents a derivative of • bm3 = 858,9 and adaptive PID models rad/s again at the 4th time
Kp, Ki, Kd which is without load second. With block diagram
capable of adjusting to the simulink as below:
change of plant parameters Fig.5 is a simulation model With input voltage of 12
following the reference or the first of a simulink volts. With gamma P,
model. plant in a mathlab gamma I and gamma D of
application, fig.6 of the 0.0502, 58, 0.000099 on
system and factory Adaptive PID, and on
B. Model Reference functions modeled into the Conventional PID kp, ki
first-order transfer function and kd of 0.01128, 1.3428,
The Reference and then estimating it into a
Model serves to provide an 0.0617743702. Getting
third-order function in results as follows:
ideal response of the mathlab with a sampling
adaptive control system to
time of 0.03 seconds and Figure 11. Conventional
the reference input. the
reference model is found the model reference is PID and adaptive PID
from the first-order model found on (39). models when inputs are
of the plant and then increased to 150 rad / s at
identifies the system and the 4th time second
estimates it into the three-
order transfer function.

Figure 9. Output
conventional PID and
adaptive PID without load

Figure 5. First-order model

of the plant
Figure 7. Output first-order
model of the plant and Figure 12. Output
reference model conventional PID and
In the graph above can be adaptive PID when inputs
seen that the reference are increased to 150 rad / s
model output graph at the 4th time second
resembles the output graph
of the desired model 1st
order plant.

V. TEST
In this chapter perform 3
tests as follows:
Figure 10. The value of Kp,
Figure 6. Step create a Ki and Kd without load
reference model
 Referensi PID PID
Performance
Model Adaptive Convesional

T(second) 0.043 0.091 0.953

ts(second) 0.867 0.972 4.411

Recovery
- 0.073 2.418
time(second)

Error(IAE) - 17.03 141.7

Error(ISE) - 367.7 1.239e4

ESS(%) 0.464 0.054 0.0046

Figure 13. The value of Kp, Figure 16. Output In the above table, the Management, Director
Ki and Kd when inputs are conventional PID and result of T, ts on PID (Academics) and Principal
increased to 150 rad / s at adaptive PID with added adaptive is smaller and for supporting us with the
the 4th time second load error using IAE and ISE get infrastructure to carry out
smaller result than the research work.
With the above graphics it conventional PID to
is found that Adaptive PID approach reference model. REFERENCES
and also conventional PID
follow the reference model [1] Najaf Abad Branch Esfahan.
VI. CONCLUSION
although the  M increase (2010). State space modeling
of 150 rad/s. In this paper and eigenvalue analysis of
the permanent magnet dc
simulates the comparison motor drive system. IEE.
3) Testing with added between the adaptive pid [2] Y. Dharshan, K. Srinivasan,
loads and the conventional pid B. Sharmila.
In the against the reference model (2017)Optimization
experiment a) carried out we want. in the simulation Technique for Networked
the loadless test and in this Control DC Motor with
results get results like the Network-Induced Delays and
experiment added the load table below. Noise.iee.
of 42.37 Nm. With block
diagram simulink as below: Concludes that the
Figure 17. The value of Kp,
best performance to
Ki and Kd with added load
regulate the speed of a dc
motor is the adaptive PID
With the added load there
with a graph having ts =
will be a  M drop to 0.972 s and T = 0.091 s
145.513 rad/s after the approaching the reference
voltage drops the response model graph. and get error
will return to the setpoint or dihitunga using IAE and
the steady response will ISE method with the result
return both the adaptive of calculation of IAE =
PID and the conventional 17.03 and
Figure 14. Conventional PID. ISE = 367.7. And dc motor
PID and adaptive PID which is given load of
models with load -42.37 Nm system will drop
B. Tables
voltage then with adaptive
Obtained T, ts, pid better to stabilize back
error and also recovery so that graph back steady or
time in the above loadless reach setpoint or called
simulation. Calculate errors recovery time equal to =
with two different methods 0.073 s.
ie with IAE and ISE and
get performance results as ACKNOWLEDGMENT
shown table below:
This research was supported
Figure 15. Model of plant and carried out at
with added load Department of Electronics
RESULT TABLE Engineering, Brawijaya
With added load and get Univerity Engineering
Result of simulation College, Malang. We would
graph simulation result like PID adaptive and PID
following: Convenional like to thank our