4th International Conference on Advances in Control and

Optimization
4th of Dynamical
International ConferenceSystems
on Advances in Control
Available onlineand
at www.sciencedirect.com
4th International Conference on Advances in Control and
February 1-5, of
Optimization 2016. NIT Tiruchirappalli,
Dynamical Systems India
Optimization of Dynamical Systems
February 1-5, 2016. NIT Tiruchirappalli, India
February 1-5, 2016. NIT Tiruchirappalli, India
ScienceDirect
IFAC-PapersOnLine 49-1 (2016) 623–628
A Study of H infinity and H2 synthesis for Active Vibration Control
A
A Study
Study of
of H
H infinity and
and H2
infinityPaavni H2 synthesis
*, Deepikafor
synthesis
Shukla for Active
Active Vibration
Ghodki**, Vibration Control
Control
Paavni
Narayan Suresh Shukla
Shukla *,
*, Deepika
PaavniManjarekar***, Ghodki**,
Pravin
Deepika M. Singru****
Ghodki**,
Narayan 
Narayan Suresh Manjarekar***, Pravin M.
Suresh Manjarekar***, Pravin M. Singru****
Singru****


*BITS Pilani Goa Campus, Goa-403726
India (Mob No.-+91*BITS
*BITS Pilani
Pilani Goa
9763555947, Campus,
Campus, Goa-403726
Goae-mail:[email protected])
Goa-403726
India (Mob
India (Mob **No.-+91 9763555947,
BITS Pilani
No.-+91 e-mail:[email protected])
Goa Campus,
9763555947, Goa-403726 (e-mail:
e-mail:[email protected])
**
** BITS
BITS Pilani Goa
Goa Campus,
Campus, Goa-403726
Goa-403726 (e-mail:
[email protected])
Pilani (e-mail:
[email protected])
*** Electrical, Electronics and Instrumentation Department, BITS Pilani Goa Campus,
[email protected])
***
*** Electrical, Electronics
Electronics and
Electrical,Goa-403726, Instrumentation
India,
and Department,
Department, BITS
BITS Pilani
Pilani Goa
(e-mail: [email protected])
Instrumentation Goa Campus,
Campus,
Goa-403726,
**** India,
Mechanical (e-mail: [email protected])
Department, BITS Pilani Goa Campus,
Goa-403726, India, (e-mail: [email protected])
****
**** Mechanical
Goa-403726, Department,
India, (e-mail:
Mechanical BITS
BITS Pilani
Pilani Goa
Goa Campus,
[email protected])
Department, Campus,
Goa-403726, India, (e-mail: [email protected])
Goa-403726, India, (e-mail: [email protected])
Abstract: Minimization of car vibrations has been a topic of research over the years and many control
Abstract: Minimization
strategies have of car vibrations
been developed. Therefore, hashas been
there a topic
needs of
to be of research over
a method the years andcontrol
many control
Abstract: Minimization of car vibrations been a topic researchfor active
over the vibration
years and many in cars.
control
strategies
This paper have beentwo
presents developed.
methods Therefore,
for therethe
designing needs to be afor
controller method
a for active
quarter car vibration
model which control
would in cars.
help in
strategies have been developed. Therefore, there needs to be a method for active vibration control in cars.
This paper
minimizing presents
car two methods
vibrations and for designing
increasing the controller
passenger for
comfort a quarter
subjectedcar model
to road which would
disturbances. help in
The
This paper presents two methods for designing the controller for a quarter car model which would help in
minimizing
methodologies car vibrations and increasing passenger comfort subjected to road disturbances. The
minimizing carused are H-infinity
vibrations and H-2 optimal
and increasing control.
passenger The subjected
comfort main aim of to this
roadarticle is to design
disturbances. Thea
methodologies
controller used are response
to control H-infinity and H-2 optimal control. The main aim of this is article is to design
modelaa
methodologies used theare H-infinity ofand
a quarter car model
H-2 optimal to various
control. road inputs.
The main aim ofThis a non-linear
this article is to design
controller
with to control
2 degrees the response
of freedom. of a quarter
It is linearized andcar model toasvarious road inputs. This is a non-linear model
controller to control the response of a quarter carmodelled a disturbance
model to various rejection
road inputs. (regulatory)
This is a non-linear problem
model
with 2 degrees
wherein the of freedom.
effect of It is linearized
disturbances are and modelled
minimised. as weighing
Different a disturbance rejection
functions have(regulatory)
been problem
considered to
with 2 degrees of freedom. It is linearized and modelled as a disturbance rejection (regulatory) problem
wherein
take the the effect
impact of of disturbances
road disturbances are minimised.
into account. Different
The resultsweighing
are functions
compared with haveProportional
the been considered to
Integral
wherein the effect of disturbances are minimised. Different weighing functions have been considered to
take the impact
Derivative (PID) ofcontroller
road disturbances
designed into account.
using the Thesystem
same results parameters
are compared and with the Proportional
performance Integral
specifications.
take the impact of road disturbances into account. The results are compared with the Proportional Integral
Derivative
The behaviour (PID) controller designed using the samesystem
system parameters and performance specifications.
Derivative (PID)ofcontroller
the opendesigned
loop andusingclosed
theloop
same systemisparameters
observed and
by examining
performance thespecifications.
impulse and
The behaviour
frequency of
responses the
in open
MATLAB. loop and
A closed
detailed loop system
discussion is is
done observed
with the by examining
choice of the
weighing impulse and
functions. It
The behaviour of the open loop and closed loop system is observed by examining the impulse and
frequency
is found thatresponses in MATLAB.
the disturbance A
rejectiondetailed discussion
techniques is
suggested done with
in this the choice of weighing functions. It
frequency responses in MATLAB. A detailed discussion is done withpaper help in
the choice of improving the system
weighing functions. It
is found that the disturbance rejection techniques suggested in this paper help in improving the system
response
is found that the disturbance rejection techniques suggested in this paper help in improving the system
response
© 2016, IFAC
response
Keywords: (International
Active VibrationFederation of AutomaticSystem,
Control, Suspension Control)Disturbance
Hosting by Rejection,
Elsevier Ltd.Quarter
All rights
Carreserved.
Model, H
Keywords: Active Vibration Control, Suspension System, Disturbance Rejection, Quarter Car Model,
Keywords:
infinity, H2 Active Vibration
infinity, PID Control,
Controllers Suspension System, Disturbance Rejection, Quarter Car Model, H H
infinity, H2 infinity, PID Controllers
infinity, H2 infinity, PID Controllers 
 The H  control technique discusses about minimizing the
1. INTRODUCTION 
1. The H
H  control
singulartechnique discusses
functionaboutover minimizing the
An automobile suspension 1. INTRODUCTION
INTRODUCTION
system is one which provides a
maximum
The control value of the
technique discusses about the Right Hand
minimizing the
maximum
Plane(RHP) singular
defined valueby of the
Re(s)>0.function
On
maximum singular value of the function over the Right Hand over
the the
other Right
hand,HandH2
An
An automobile suspension system
smoothautomobile
ride by suspension
separating is
is one
the vehicle
system one which
body from
which provides
roadaa
provides Plane(RHP)
optimal
Plane(RHP) defined
control
defined by
by Re(s)>0.
technique is one
Re(s)>0. On
On the which
the other
otherdeals hand,
hand,withH
H22
smooth
smooth ride by separating the vehicle body from road
ride
irregularities by
and separating
bumps. the
However, vehicle
a body
passive from
suspension
road optimal
minimization control
optimal control technique
of energy(or
techniquepower) is one
is one which
over which
the entire deals
dealsrangewith
of
with
irregularities
system is unable
irregularities and
and tobumps.
provideHowever,
bumps. this. Active
However, aa passive
suspension
passive suspension
systems
suspension minimization
minimizationAof
frequencies. energy(or
ofnumber
energy(or power)
of studies
power) haveover
over beenthe entire
thedone
entirebothrange
on Hof
range of∞
system
on
system is
is unable
the other hand to
unable have
to provide this.
this. Active
force actuators
provide suspension
designed
Active systems
separately
suspension systemsfor frequencies.
and H2 control
frequencies. A
A number of
of studies
techniques.
number have
have been
Studies
studies like done
been thoseboth
done both on
on H
done Hby∞
on
on the other hand have force actuators designed separately for
the
improvingother hand
ride have
comfort force
by actuators
minimizing designed
the separately
effect of road
for ∞
and
and H22 control
H control techniques.
(Thompson,1980) show howStudies
techniques. H∞ like
Studies those
those done
is preferable
like doneoverby
by
improving
improving ride
disturbances comfort
(Du,2012
ride comfort andby minimizing
minimizing the
byRao,2013). the effect
effect of of road
road (Thompson,1980)
LQR(Linear Quadratic
(Thompson,1980) show how
how H
showRegulator) is
is preferable
H∞∞ as system performance
preferable over
over
disturbances
disturbances (Du,2012
(Du,2012 and and Rao,2013).
Rao,2013). LQR(Linear
specifications Quadratic
are directly Regulator)
incorporated
LQR(Linear Quadratic Regulator) as system performance as system
into the performance
optimal cost
The analysis of road disturbances on car as well as the specifications
The analysis of specifications are directly incorporated into the optimal cost
function. Many are directly
such similar incorporated
studies into
compare theLQRoptimalwith H∞
cost
passenger
The analysis of road
is done using disturbances
road disturbances on
on car
a quarter car model. carTheas well
well as
as linear the
model
as the function. Many
passenger
we have is
takendone
for using
analysisa quarter
is done car
bymodel. The
(Anilkumar, linear
2013)model
and
control Many suchsuch similar
function.technique(Shirdel similarand studies compare
compareInLQR
Nagarkar).
studies LQRa studywith
with HHby∞∞
passenger is done using a quarter car model. The linear model control
we control technique(Shirdel
(Abdellahi,2000) importance
technique(Shirdel and
andofNagarkar).
tuning H∞In
Nagarkar). aa study
study by
In controller by
we have
have taken
(Kousik, 2014)for
taken analysis
forand it has
analysis is
is done
two by
done (Anilkumar,
translational
by (Anilkumar,degrees 2013)
2013) and of
and (Abdellahi,2000) importance of
(Kousik, choosing appropriate
(Abdellahi,2000) importanceweighing of tuning
functions
tuning H∞∞ iscontroller
H by
illustrated.
controller by
(Kousik, 2014) and it has two translational degrees of
freedom (2 2014)
DOF). and
The it has
design oftwo translational
force actuators in degrees
our case is
of choosing appropriate
(Kaleemullah,2011) has weighing
compared functions
H with is
LQR illustrated.
and fuzzy
freedom choosing appropriate weighing functions ∞ is illustrated.
freedom (2 DOF). The design of force actuators in our case is
done (2
using DOF).
H The
infinity design
and H2of force
optimal actuators
control in our case
techniques. is (Kaleemullah,2011)
control and it has compared
done using
thenH infinity logic
(Kaleemullah,2011) is found
has compared that HHH∞∞∞ with
with LQR
LQRinand
is best and fuzzy
terms of
fuzzy
This
done is using infinity and
H compared andto H2
H2the optimal control
conventional
optimal techniques.
controlProportional
techniques. logic
This logic control and it is found that H∞ is best in terms of
control
robustness and it
specification. is found
Not muchthat H ∞ is best
research has in
been terms
done on
of
This is
Integralis then
then compared
Derivative
compared(PID)to the
the conventional
to controller. A study
conventional Proportional
done by
Proportional robustness
2 optimalspecification. Not
Integral H
robustness control technique.
specification. Not much
muchBut,research
research has
has been
it has been beenshowndone on
donethat
on
Integral Derivative (PID) controller. A study done
(Kruczek) Derivative
shows how (PID) controller.
passenger comfort A study
and ride handling
done by
by H 2 optimal
LQR control
is similar to H technique.
2. Inspite But, of thisit has been shown
comparison that
between
(Kruczek)
are two shows
contradictory how passenger
requirements comfort
which a and ride
controller handling
like H H 2 optimal control technique. But, it has been shown that
(Kruczek) shows how passenger comfort and ride handling∞ LQR
H  andis similar
mixed H to H . Inspite of this comparison between
are
can’ttwo contradictory
handle at once. requirements
Therefore, two which aa controller
of these controllerslike H LQR is similar to H/H 2 has been done.
2. 2 Inspite of this comparison between
are two contradictory requirements which controller like are
H∞∞ H H /
can’t
designedhandle
for at
eachonce.
of Therefore,
the techniques two of these
mentioned
can’t handle at once. Therefore, two of these controllers are controllers
above, one are
for H  and mixed H  /H22 has
and mixed H has been
been done.
done.
designed
designed for
improving each
each of
forPassengerof the techniques
comfort
the techniquesand mentioned
the other for
mentioned above,
above, one
one for
optimizingfor All these studies have motivated us to compare between the
improving
Suspension Passenger
Deflection. comfort and the other for optimizing All
All these
two control
these studies have
have motivated
strategies-
studies H  and H
motivated us to
to compare
us2.Also, between
PID controller
compare between has the
the
improving Passenger comfort and the other for optimizing
Suspension
Suspension Deflection.
Deflection. two control strategies- H and H .Also, PID controller has
two control strategies- H  and H2.Also, PID controller has
2

2405-8963 © 2016, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.
Copyright © 2016 IFAC 623
Peer review under responsibility of International Federation of Automatic Control.
Copyright © 2016 IFAC
10.1016/j.ifacol.2016.03.125 623
Copyright © 2016 IFAC 623
IFAC ACODS 2016
624 Paavni Shukla et al. / IFAC-PapersOnLine 49-1 (2016) 623–628
February 1-5, 2016. NIT Tiruchirappalli, India

been used for benchmarking purpose. The paper consists of x 2  (1 / m p )(k p ( x1  x3 )  j p ( x1  x3 ) 3  c p ( x2  x4 )
the following sections. Section 2 deals with the modeling of
Quarter Car model. Section 3 tells about designing of  b p ( x 2  x4 ) 3 ).
controllers using the above mentioned techniques. Section 4
(4)
shows the result of the simulation of the controllers so
formed. Finally a conclusion is drawn about the usage of both x3  x4 . (5)
the controllers after a brief discussion. x 4  (1 / mc )(k p ( x1  x3 )  j p ( x1  x3 ) 3  c p ( x 2  x 4 )
2. MATHEMATICAL MODELLING OF QUARTER CAR
MODEL  b p ( x 2  x 4 ) 3  k c ( x3  r )  j c ( x3  r ) 3  cc ( x 4  r)
A Quarter Car model is a 2DOF system having two  bc ( x 4  r) 3 ).
translational degrees of freedom. It is used to generate the (6)
represent suspension system. State space model of quarter car In order to find the stable fixed points, we observe the nature
has been taken from (Anilkumar, 2013) for reference. of Eigen values when x1  x 2  x3  x 4 =0 and road input
2.1 Open- Loop and Closed- Loop System r= =0. As expected, it is found that the only stable fixed
point is x1  x2  x3  x4 =0.The system is controlled in
such a way so as to bring it back to this stable set of points
when a disturbance is introduced.

In a closed loop system, there’s an additional damper or a

force actuator. The designed controller would measure the
displacements of passenger and would produce a force Fs on
mass mc instantaneously so that some vibration matrices are
minimized. With the addition of controller force (6) is
modified as:
x 4  (1 / mc )(k p ( x1  x3 )  j p ( x1  x3 ) 3  c p ( x 2  x 4 )
a) b)
 b p ( x 2  x 4 ) 3  k c ( x3  r )  j c ( x3  r ) 3  cc ( x 4  r)
Figure 1: Coupled Passenger Quarter Car Model for a) open
loop and b) closed loop case. Here subscript ‘c’ represents car  bc ( x 4  r) 3  Fs ).
and subscript ‘p’ represents passenger. (7)
Table 1 : Model Parameters A plant or a system is represented as the following

Symbol Description Value Unit

mp passenger mass 60 Kg
kp linear stiffness coefficient 25000 N/m
jp cubic stiffness coefficient -2500 N/m3
linear damping
cp 200 Ns/m Figure 2:Diagram of a system with controller
coefficient
bp cubic damping coefficient -10 Ns3/m3 e  w  P11 ( s) P12 ( s)  w
mc car mass 240 Kg  y   P( s)  u    P ( s) P ( s)  u . (8)
jc cubic stiffness coefficient -300000 N/m3      21 22  
kc linear stiffness coefficient 160000 N/m u  K (s) y. (9)
cc linear damping coefficient 250 Ns/m
e  F1 ( P, K )w. (10)
bc cubic damping coefficient -25 Ns3/m3
From Figure 1a), the following equations of motion can be where F1 is the lower linear fractional transformation given
written 1
by- F1 ( P, K )  P11  P12 K (1  P22 K ) P21. (11)
m p x p  k p ( x p  xc )  j p ( x p  xc ) 3  c p ( x p  x c )  b p ( x p  x c ) 3  0.
Table 2: Parameters of System with controller
(1) w Exogenous input-Commands, disturbances, noise
mcxc  k p ( x p  xc )  kc ( xc  r )  j p ( x p  xc )  j c ( xc  r )3 
3
e Regulated output-error tracking, actuator signals
c p ( x p  xc )  bp ( x p  xc )3  cc ( xc  r)  bc ( xc  r)3  0. (2) y Sensor output-Measured plant output, measured
disturbances
u Controlled input-Commands, measured plant outputs
In order to reduce the above equations into a convenient form and disturbances
for analysis, we take x1  x p , x2  x p , x3  xc , x4  xc , P Plant-state space equations or matrix of the system or
plant
x1  x2 . (3) K Controller

624
IFAC ACODS 2016
Paavni Shukla et al. / IFAC-PapersOnLine 49-1 (2016) 623–628
February 1-5, 2016. NIT Tiruchirappalli, India 625

Linearising equation 7, we can represent our plant in state-

space form as below-
x (t )  Ax(t )  Bu(t ). (12)
y(t )  Cx(t )  Du(t ). (13)
 0 1 0 0 
 
A  (  k p / m p ) ( c p / m p ) (k p / m p ) (c p / m p ) 
 (k p / mc ) (c p / mc ) ((k p  kc ) / mc ) ((c p  cc ) / mc )

0 0 0   1 0 0 0  0 0 0
0 0   kp  cp kp cp  0
 0
  0 0 Figure 3: Plant or the quarter car model with the weighting
B 0 0 0  , C   mp mp mp m p , D  
 1 0 0 functions
 kc cc  1  0 0 1 0 
 
   0  0 0 1 The figure above shows the connection diagram of the
 mc mc mc   0 0 0 
system. From the above diagram, we can see that if
From this, we can conclude that the outputs are- Sensitivity of the controller=S = (1  GK ) 1 , Then
y1 (t )  x1 (t ); y2 (t )  x2 (t ); y1  Wx1 y. (17)
y3 (t )  x3 (t )  r; y4 (t )  Fs y2  Wact u. (18)
Solving the state space matrix (8 and 9), by Laplace
transformation, we get the plant matrix as the following- y1
1
 Wx1Wref S . (19)
G  D  C(s  A) B. (14) D1
Where G is the plant and its components are y2
P11 , P12 , P21 and P22 .  WactWref KS . (20)
D1
y3
3. DESIGN OF CONTROLLERS  Wx3rWref S. (21)
D1
3.1 PID Controller Following is the optimal choice of each of these weighing
functions for simulation purposes.
For designing of PID Controller, Tune option present in
 Wref- This is connected to the input disturbance
Simulink has been chosen. Two cases as considered by
signal d1. In order to reduce the effect of high
(Kousik,2014) have been taken. In the first case, Passenger
frequency road disturbance on the system, it is
acceleration( x 2 ) has been given as feedback and in the chosen to be a low pass filter. In our case
second case, Suspension deflection(x3-r) has been used. 16
Thus both these controllers have different coefficients which Wref  .
affect their performances. s  200
 Wn-This is present at the control input of the
3.1 H∞ Control system. This is taken to be a constant of small
H∞ norm of transfer function T(s) is defined as- value to minimize the sensor noise. Here, we
T  maxR  [T ( j). have taken it to be Wn  0.01
 (15)
 Wact- This is the performance weighting function
Which means that it is the maximum singular value or the and is situated at the controller output. Since it
greatest magnitude of the transfer function over all the acts on the force generated by the controller, it is
frequencies. In our case with the plant P and input-output kept a high pass filter. At high frequencies as
map F1(P,K)(or Tzw ) , H  algorithm minimises the y2
sensitivity S  0 , 0 This ensures
normover all realisable controllers K. d1
minimization of actuator force at both high and
3.2 H2 Control low frequencies. In our case,
H2 norm of transfer function T(s) is defined as 0.3( s  50)
Wact 
1
 1500( s  500)
T (s) 2   T ( j )T ( j ) d .
T
(16)  Wx1 and Wx3r-These two are the robustness
2  weighing functions connected to the first and
Here T(s) is the transfer function of the whole system third output. Since we don’t want high frequency
including the plant P and the controller K along with user components of passenger acceleration or
defined weighing functions. H2 norm is a measure of energy suspension deflection, these are considered to be
(or power) of the system and we try to minimise this energy low pass. Both these things can’t be controlled
for all the frequencies. simultaneously, hence there is a trade-off

625
IFAC ACODS 2016
626 Paavni Shukla et al. / IFAC-PapersOnLine 49-1 (2016) 623–628
February 1-5, 2016. NIT Tiruchirappalli, India

between the two. We have defined a factor β to

take this into account. So
kc 200
Wx1  
mc ( s  200)
kc 200
Wx 3r  (1   )
mc  m p ( s  200 )
4. RESULTS OF SIMULATION
Two cases are taken. For Controller A which minimizes
passenger acceleration, β=1 and for Controller B which
minimizes Suspension Deflection β=0.These two controllers
are made for both H∞ and H2 norms and the results are
compared with the two cases of PID controllers. Two inputs,
one sinusoidal and the other pulse in nature are taken. Figure 5: Passenger acceleration and Suspension Deflection
with and without controller A for minimising Passenger
Below is the profile of the road input which is the disturbance acceleration (β=1) in case of sinusoidal input
‘ r’ fed as exogenous input to the plant. The height of bump
and pothole is 0.1m each and last for a period of one second.
Hence we can estimate the reaction of the car and passenger
while encountering a bump and a pothole.

Figure 4: Input Profile- The sinusoidal bump in the design

represents a speed breaker and the window function can be
thought of as an inverted pothole.
4.1 Open Loop
Initially, the response of the system ie. passenger and car’s
displacement and velocities are observed without the
controller. From the graphs below, it can be observed that the
vibrations die out 3-4 seconds after the disturbance is
encountered.

Figure 6: Passenger acceleration and Suspension Deflection

with and without controller A for minimising Passenger
acceleration (β=1) in case of pulse input

626
IFAC ACODS 2016
Paavni Shukla et al. / IFAC-PapersOnLine 49-1 (2016) 623–628
February 1-5, 2016. NIT Tiruchirappalli, India 627

5. DISCUSSION

Table 2 Comparison of Controller A and Controller B for

the two inputs

Sinusoid Controller A (Case1) Controller B (Case2)

al Input

Peak, % Settling Peak, % Settling

decrease Time, % decrease Time,
decrease %
decreas
e

Passenger Acceleration

Open 12.5 m/s2 4.28s 12.6 m/s2 3.46 s

Loop

H2 3.5 m/s2, 72% 1.65s, 11.5 m/s2 1.32 s,

61.45% , 8.7% 62%

H- 6 m/s2, 1.35s, 10.8 m/s2 1.64 s,

infinity 52% 68.46% , 14.3% 53%

PID 0, 100% - 5 m/s2, 2.98s,

Figure 7: Passenger acceleration and Suspension Deflection
60.32% 13.87%
with and without controller B for minimising Suspension
deflection (β=0) in case of sinusoidal input
Suspension Deflection

Open -0.01 m 4.43 s -0.01 m 3.89 s

Loop

H2 -0.02 m, 100% 1.09 s, -0.01 m, 1.41 s,

75% 0% 64%

H- -0.02 m, 100% 1.54 s, -0.01 m, 1.64 s,

infinity 65% 0% 58%

PID -0.09 m, 800% 1.64 s, -0.01 m, 2.74 s,

62.98% 0% 29.56%

Pulse Controller A Controller B

Input

Passenger Acceleration

Open 66 m/s2 3.97 s 66 m/s2 3.19 s

Loop

H2 18.5 m/s2 , 1.36 s, 59 m/s2, 1.4 s,

72% 65.74% 10.61% 56%

H- 44 m/s2 , 89% 1.38s, 52 m/s2, 1.59 s,

Figure 8: Passenger acceleration and Suspension Deflection
infinity 65.24% 21.21% 50%
with and without controller B for minimising Suspension
deflection (β=0) in case of pulse input

627
IFAC ACODS 2016
628 Paavni Shukla et al. / IFAC-PapersOnLine 49-1 (2016) 623–628
February 1-5, 2016. NIT Tiruchirappalli, India

Ales Kruczek, Antonin Stribrsky, “H∞ Control of

PID 0.2 m/s2, 1.23s, 33.4m/s24 2.87s,
Automotive Active Suspension with linear motor”,
99.6% 69.02% 9.39%, 10.03%
A.H. Shirdel ,E. Gatavi, Z. Hashemiyan, “Comparison of H-
∞ and optimized-LQR controller in active suspension
Suspension Deflection system”, Second International Conference on
Computational Intelligence, Modelling and Simulation,
Open Loop 0.10 m 1.16s 0.09 m 3.12 s pp. 241-246
Allen Anilkumar and Pravin M. Singru(2013), "Seat Design
H2 0.09 m, 1.04s, 0.09 m, 2.73s, and Control of Passenger Vibrations in a Quarter Car
10% 10.34% 0% 12.5% Model Using h-∞ Controller", 11th International
Conference on Recent Advances in Structural Dynamics
H- infinity 0.09 m, 1.14s, 0.09 m, 2.44s, E. Abdellahi, D. Mehdi, M. M’Saad(2000), “On the Design
10% 1.72% 0% 21.79% of Active Suspension System by H∞, and mixed
H2/H∞,: An LMI Approach”, American Control
PID 0.099 m, 1s, 0.09m, 1.04s, Conference, Chicago, pp. 4041-4045
1% 13.79% 0% 66.67% Haiping Du, Weihua Li, and Nong Zhang(2012), “Integrated
Seat and Suspension Control for a Quarter Car With
The following observations can be made from the above Driver Model”, IEEE Transactions on Vehicular
analysis- Technology, pp. 3893-3908
Kaleemullah M, Faris W.F, Hasbullah F(2011), “Design of
 Oscillations of higher amplitude are generated when Robust H∞, Fuzzy and LQR Controller for Active
a pothole is introduced as compared to a speed Suspension of a Quarter Car Model”, 4th International
breaker. Conference on Mechatronics (ICOM), Kuala Lumpur,
 There is a significant decrease in peak values of Malaysia
acceleration in Controller A as compared to M. P. Nagarkar, Dr. G.J. VikhePatil, “Performance Analysis
Controller B. Therefore, Controller A is found to be of Quarter Car Active Suspension System: LQR and H∞
better than Controller B for all the three techniques- Control Strategies.”, ICCCNT'12, Coimbatore, India,
H∞, H2 and PID. 26th-28th July 2012
 Although PID controller is successful in minimising Peter M. Thompson and Richard Y. Chiang(1980), “H∞
passenger acceleration, in terms of Suspension robust control synthesis for a fighter performing a
Deflection, H∞ and H2 controllers generally give coordinated bank turn”, Conference on Decision and
more controlled results. Control, pp. 3362-3366
 It is observed that H2 norm aims to minimise all P. Kousik, Pravin M Singru and Narayan Manjrekar(2014),
singular values at all frequencies by minimising the ”Comparing PID and H-∞ Controllers on a 2-DoF
energy of the system. On the other hand, H∞ norm Nonlinear Quarter Car Suspension System” Journal of
minimises the maximum singular value at the Vibro-Engineering
highest frequency. Tinnavelli Ramamohan Rao, Punjala Anusha(2013), “Active
 This implies that H2 control gives better results as Suspension System of a 3 DOF Quarter Car Using Fuzzy
compared to H∞ when reducing the disturbances in Logic Control for Ride Comfort”, 2013 International
all frequencies in general. Though, there are Conference on Control, Automation, Robotics and
situations where H∞ performs better. A choice has Embedded Systems
to be made by the designer according to the situation
and preference he has. Moreover, an approach of
mixed H2/H∞ synthesis with constraints on pole
placement can be used. The resulting norm would
contain two terms along with the weighting
functions for both of them (using hinfmix function
in MATLAB).
6. CONCLUSIONS
A detailed analysis has been done between H2 and H∞
control techniques. Using PID as a benchmark controller has
helped us to test the performance and robustness of these
controllers. The weighting functions play a significant role in
deciding the performance of the controllers. These controllers
prove to be effective in reducing road disturbances and
tackling different inputs. As part of the future study, a mixed
H2/H∞ synthesis can be done which will combine both the
norms and various cases may be taken.
REFERENCES

628