orapter 19
Non-Linear
Control Systems
We have alrendy introduced linear control systems and studied them from their pole-zero configuration,
hich is an extremely powerful tool both in. frequency and time-domain analysis and synthesis. Unfor-
tunately, systems that exist in nature are non-linear, Hence, the superposition principle which is well
valid in linear systems js totally invalid in non-linear systems, There are control systems which do not
have any linearized version and itis very difficult to linearize these control system. In some cases, it has
been found that the non-linearities such as dead zone and hysteresis and combination of these are unde-
sinble and unavoidable, The aim of this chapter is to discuss the methods to study non-linear systems
aswell as their stability, :
19.1 The State Space Approach
Figure 19.1 shows the block diagram representation of a class of non-linear control systems. The plant
‘dynamics is represented by the state ve
/ector x. In Fig, 19.1, ris the reference, w is plant input ahd c is the
controlled output vector,
tr Controller uv
Fig. 19.1, Non-linear control system.
The mathematical models to represent the plant and controller are given below.
Plant
d= f(x,u) (19.1)
and
car (19.2)
Controtier
u=g(cr)
(19.3)Qe
798 | Control Systems
In Eqs. (19.1) and (19.2), the vectors are in general non-linear functions, and we have
controller does not contain any energy storing elements. Substituting Eq. (19.2) in Eq, (lo: ioe 3) ete
We ger
= g(Cr)
Substituting Eq. (19.4) in Eq. (19.1), we get (194
= £(,8(Cx,r)) = Fr) ne
If the input vector r is assumed to be consiant, Eq. (19.5) can be written as »
b= F(nr) = Fx
x= F(x,r)= F(x) (199
Lot us take a single output system as shown in Fig. 19.2 to obtain the state space equations forty
given system.
+V py
ic ie a
. a(s+2)
aE
Fig. 19.2 Single output system
For plant we can write
1
= VO
This gives the differential equation w=€+2é. Putting c=x, and x,
=u-2é=u-2x,.
In state space form we can write the plant differential equations as follows:
= fu)=x,
and
= f,(xu)=-2x, +u
The maximum effort of the controller is +U,, to -U, when error é = r-c is positive and negative,
respectively. Mathematically we can write
roc rx,
Ir-e] "~ [r=x]
U,sen(r-2,)
where the function sgn(r-x,) has a value + 1 when (r — x,) is positive and it has a negative value wheo
(r-x,) is negative.
In state space form wé can write the controller differential equations as follows:
=F(x,r)=x,
F,(x,r) = -2x, +U, sgn(r—,)Non-Linear Control Systems | 799
phase Plane Method
e hserved that non-linear system may be simplified to a non-linear second-order system wit
oss 220 aproximations. We can apply very easily a ‘graphical representation of the phase trajectory
foe d-ordr system, Isoclnes in a phase plane area very general and convenient graphical
i aoe ong other existing methods. The main disadvantage in the phase plane technique is that the
gars aene ea explicitly inthe trajectory. Let us discuss a graphical method for determining the
does nO for a second-order system represented by .
rajectory ;
e E+ fi(c,e+ F,(de=r (19.7)
ye f(od)éand flGAE are functions of c and é respectively
ere i
=c and x, =6, we get form Eq. (19.7)
putting %,
(19.8)
wd
1 ~fQ%)% —AGp)% , (19.9)
From Eqs. (19.8) and (19.9) we have
3 x
fie %) Lex) tte Z (19.10)
44 :
Equation (19.10) gives the slope of the trajectory at any point (x, x,). Let us take
ax, 4
A, 19.11)
&, ™ (19.11)
where m is constant. We can write
F(x,x,)=m
x ae
FQ x)= =fiG%)- AOE (19.12)
2M
Any trajectory can cross the curve described by Eq, (19.12) and this curve is known as an isocline.
Let us consider a second-order linear system as shown in Fig. 19.3 with input r(¢)=2u(#) and
output c(t) having the following initial conditions:
(0) =I and (0) = 0
From Fig. 19,2, we can write 5 -
(8? +25+5)C(s) = 5R(S) z y 5(6+2) &
Taking inverse Laplace transform, we get
64264+5e=5r210 (19.13) Fig. 19.3 Second-order linear systemFo
800 | Control systems
Putting x, =c and x, = é, comparing Eq. (19.13) with Eq. (19.7) we get
Alem) =2
aGiv%)= 5
As per Eq. (19.13), we get
x10
m=-2-5++—
aE
The family of isoclines will have the following equation:
SL ply
nt med
which describes a family of straight lines. The slope of these lines is given by
These lines intersect the x axis at a point having
the following value at the x coordinate:
ie, 22.
The common point of all isoclines is (2, 0). If
m= 0, the equation for the isoclines will be
aa-Sa45
Figure 19.4 shows the isoclines for different
values of m.
To get the phase plane trajectory foranon-linear
system, let us consider Fig. 19.5 where inductance
Land capacitance C are linear but resistance R is
non-linear. Let the current characteristics of the
non-linear resistance be given by Fig. 19.4 Isoclines of Fig. 19.3
i=-AV+BV?
From Fig. 19.5, we get L v i
1 av
lV -DarcT—av +B =0 (19.16)
Differentiating Eq. (19.16) wars. 1, we get
av av
—— + —[3LBV? — =
wat Fb ALJ+V =E
i Fig. 19.5 A systemNon-Linear Control Systems | 01
(19.17)
wd
#.(4 (ae 49.18)
. ae \icyae
sing Eqs. (19-17) and (19.18), we get from Eq. (19.16) as
av dv
ot Oly af aa [E rer
a w 19.19)
its Brae ery
bere anale
panes
A
Form Eq, (19.19), we have
2
oT a -bY)]4V =E
the standard values a = 0.2 and b = 1, we can
Setting x,(t) = v(t) and x (D=VO) and assuming
wit the following:
x,
=020-33)-%
aS -4- 0.20-. -2)-3
“m=
The equations of isoclines are given by
42 q-5m) 3 (19.20)
For m = 0, the isoclines are given bY
5:-
802 | Control systems
| Fig. 19.6 Isoclines and trajectories for a non-linear system
Figure 19.6 shows the isoclines and trajectories for a non-linear system for different values of m. Fora
non-linear system, we get a family of trajectories. Therefore, this diagram is also called phase portrait
19.3 Describing Function Analysis
The steady-state output of a non-linear device will have the same period of output, but there is 0
guarantee that the output will be sinusoidal in spite of sinusoidal input. We know that the Fourier analysis
of the fundamental ‘wave form will lead to a fundamental component as well as several harmoniss.
It is assumed that subharmonics do not exist. If the feedback system shown in Fig. 19.7 is of low pis
nature, the harmonics will be attenuated and may be neglected to get a reasonable approximation tothe
performance of the system. There are possibilities of oscillations if the loop gain is one with a phat
shift of 180°,
+ Non-linear
Als) G(s) oa Gis)
His)
Fig. 19.7 Non-linear system-
I Non-Linear Control Systems
lex "
1 son ‘ofa nonlinear element and itis given by
a
Cd fac
"ah
isthe am
te he OUP
go the input
faxes ot
and @, is the phase shift of the fundamental harmonic component
d= A +E (4 sine B, coset) =, +57, sin(nar +4)
2 pe
JF vO cosnestd (or), B, = ok ‘y(e) sin n aot d (1)
y= (AB
ww (2)
| oven non-linearity is symmetric,
The fundamental component of the output is
y, = A,cosat +B, sinant= ¥,sin(o+9,),
where
1
[y@eosaraco amd ani ‘y(t)sin oot dct)
4 T
‘The amplitude and phase angle of ‘the fundamental component of the output are given by
Ee
“()
"Pectively.
19,
3.1 Dead-Zone Non-Linearity
Rig
ee 8 shows idealized characteristics of non
Aen input. No output can be obtained within
ratio of the fundamental harmonic component of the output to the input define
| 803
s the
Jitude of the input sinusoid, Y, is the amplitude of the fundamental harmonic compo-
of the output with
ibe the input to a non-linear element, by Fourier series the output is given by
(19.21)
(19.22)
linearity having dead-zone and its response due to
the dead-zone amplitude for inputs to a dead-zonex() =X sin ot
Fig. 19.8 Sinusoidal response of non-linearity with dead-zone linearity
From Fig. 19.8 the output y(/) for 0 < wt $ mis given by
0 for O
] 7] 20° (19.26)
4 goro| eM {ps (2) see
A
Figure 19.9 shows the plot of N/K versus =a
(19.25)
‘he dese
08
06)
xlz
04
02
“0 602 04 06 08 1.0
&
x
Fig. 19.9 Describing function for the dead-zone non-linearity
|
|
|
|
|
183.2 Saturation Non-Linearity
Fue 19.10 shows a typical input-output characteristic for the saturation non-linearity. In this case, the
Of the saturation element is proportional to the input for small input signals, the output will not
Proportionally for larger input signals and at last the output is constant for very large signals.
Figure 19.10 shows thatthe output is odd and the fundamental component contains only sine terms,
ntal harmonic component of output is given by
WO) = B, sinoWit) = yy Sin ot
x(t) = X sin at
Fig. 19.10 Sinusoidal response of non-linearity with saturation non-linearity
where
1 yosinaxd wr) = {Rosin ord (wr)
(19.27)
i
©
From Fig. 19.10, the output y(t) for 0< @r 0.
K(X-A) forazor2B
y()=4K(S-A) for B2ar>(n-B)
K(X-A) for (n-B) 2 of 2 (n-@)
0 for (n-0) 20127
Since the output is an odd term, w= D8, coset.
el
The output has also half-wave as well as quarter-wave
symmetry also, The fundamental harmonic component is
given by
»()= B, sinot
14 i
a
12 CI Y
10 na
a ‘San
N fir
6 rt
4 my
3 LH
Fig. 19.16 Describing function for the
on-off non-linearity
Fig. 19.17 Sinusoidal response of non-linearity with dead zone and saturationrrr Se i
Non-Linear Control Systems | 814
“ x
. : [xo sinaraco=2 [7 y@sinotd(ot)
4g F
: Ji KCxsinwt—A)sinord(on += ff K(S—A)sin ord(o
[2(B— 0) + (sin 2B -sin 2a.)]
he describing function is given by
0 for|X|
N _} 1-2 @+sineccoso. for A<|X|<5,B== 19.35)
(
| K x 2
| 1 2@@-a)+(sin2B-sin2a)] for|X|>S
©
Forsaturation non-linearity
A=0 and a=0
Therefore, the describing function in given by
1 : for|X|S (19.36)
Figure 19.18 shows the describing function of saturation non-linearity.
Dead-zone non-linearity In this case S—> o, B = 1/2. Therefore, the describing function is given by
0 for|X|A
Figure 19.19 shows the describing function with dead-zone non-linearity.812 | contro! systems
XN
K N
10 K
a 1.0
08
as os
04
02
oq
Poe ee
ote2e4680 X 24 6 8 10
s
x
é
Fig. 19.18 Describing function of Fig. 19.19 Describing function of
saturation non-linearity dead-zone non-linearity
For relay with dead zone In this case S = A and slope K = eo and the saturation value is |0|. Taking
a= B, the expression for N/M is given by
0 for|X|A
19.3.6 On-Off Non-Linearity with Hysteresis
Figure 19.20 shows the input-output characteristic curve of an on-off element. Let the input (4) = X'sin ot,
X> h where h is the dead zone of the on-off non-linearity with hysteresis. The output is given by
M fort, sts(E+s)
-M tor(Eo)si<(224 )
o o
4) Due to odd symmetry of the output, its fundamental contains only sine tems.
The fundamental harmonic component is given by
= (1939)
where of, = sin
where
Pipettes a
zB, =f y(Osinotd(cr) = 212" yosinard (a)
=2 fa sinavd (on =
ne :
ple S
X WXNon-Linear Control Systems | 813
output
yi) ym”
-h ym
Bn '
Beat
* ee
xt) =Xsin of
8
iv
rig. 19.20 Sinusoidal response of relay with hysteresis-type non-linearity
wave and it lags behind the input by an angle ~sin' (WX) and the describing
‘pe output is a square
fuooion for the non-linear element is given by
4M.
Nelo Z-sio
aX
‘N versus h/X.
( Jin arr instead of.
x x
Generally its convenient fo plot (HIMYN =
Figure 19.21 shows the plot Of ANIM versus HX.
{9.3.7 Relay with Dead Zone and Hysteresis
Figure 19.22 shows the input-output characteristics of @ relay with dead-zone hysteresis. The output yO)
isgiven by
9 foros wt S™
4M for 0S oF sm
yt=7° for (n-B)S OFS (n+0)
M for (ara) sos 2R-P)
9 foran-BS% <2n
1 [4 —2h/X]-
here c= sin! (A/X) and B= sitPr ae ee.
814 | Control Systems
aa 10°
20°
10 ape
— 40°
08 50"
60°
06 ~70°
hn A) -e0e
04 M -80°
02
0 02 04 06 08 1.0
Fig. 19.21 Describing function for the non-off non-linearity with hysteresis
(n+)
ot
| 2x
(2n+B)
Fig. 19.22 Sinusoidal response of relay with dead-zone and hysteresis
2p 2 ie 4Mh
4 =f YQ) cos ar d(at) = S(t cose daw) =— =F
gi aleNon-Linear Controt Systems "| 815
i ZU 0 sinex ator = 25 a sinc dwt)
= A (eosa.tcosp)
22M) Winfia yi 4-2hy
A -F
2%) 3B, _2M ay A-2hy
Xo ax x) Md Yao 1-(2) + 1-(224)
herefore, .
7 e
4 ) 3 B,
4) fA tf Be
Re (4 x | <1 4 for |X|2
4 2
19.4 Stability
The stability characteristic of a non-linear system
canbe determined using Lyapunov method. The fol-
loving definitions are important before studying the
Iyapunov method.
Letus consider a scalar function F of state variable
Bie, Fl) = Pict, xy xy. x). If the function
Phas always a positive or negative sign in a region
‘bout the origin except at the origin, the function is
‘id to be positive or negative definite. The function
Feal+a2+32 is positive definite in three-
ensional space except at the origin. The function
2x'~3x3 Sx? is negative definite in three-
oenionl space except at the origin. The function -60°
,, “i~5)+ (x, ~7) isnot positive definite because
= values at x,=5 and x,=7 also. This -90°
nis called positive semi-definite. Similarly, the
Fax 5) —3(x, —7)? is not negative us 5
‘ithas zero valuesat x, =5 and x,=7 Fig. 19.25 Describing function of backlash
ion is called negative semi-definite. non-linearity
IKI
“Nxa
818 | Control Systems
19.4.1 Lyapunov’s First Theorem
If there exists a scalar function F(x) for an autonomous system x= F(x) such F(x) is real, continuous
with continuous first order partial derivatives and
FQ)=0 :
F(x)>0. forall x#0, and
AFC) © forall x#0
dt
1m is asymptotically stable in the neighbourhood of the origin.'On the other hand, it
Then, the syste
F(2) — o, the system is stable but there is no assurance for asymptotic stability. The function F(x) is
dt
known as Lyapunov's function. Let us consider and autonomous system expressed as
Rex
5 =m %
Let us take Lyapunov’s function as
Faxtx
Here F = x? +x2 is positive definite. Therefore, F(0)=0. Now,
dF _ OF ox, OF ax,
‘dt dx, Ot Ox, at
= 2x,8, + 2x, (-m,%, — m,X,) = 23 ,¥, — Dx. — 24 MH
= 2(4, =x, — 4%) = [20-—m, )x, - 2m, x5],
For m,=1 and m,>0, we get
aF
ia
Therefore, for any value of x,,d¥/dt = 0 if x,
19.4.2 Lyapunov’s Second Method
IfA is a positive matrix, we can write
© F(@)=x7Ax
8F(x)
or
Again i= Bx. Therefore, 37 = x7 BT
3F@)
&
XT AR+XTAR
= Axt x7 At = x7 Bl Ax+ x7 AB = x! (BT A+ AB)xNon-Linear Control Systems | 819
C=-(BYA+ AB)
ut
aF@) _ -
se oO = F()=-x7Cx.
selecting arbitrarily a positive definite or semi-definite matrix for C, we can solve for C. The neces-
se sufficient condition for asymptotic stability is thatthe matrix C must be positive definite.
38 arin isd tobe postive definite if ts symmetic component is positive definite.
Now
[eras an)Lura+20)]
<_[gr( Ata), (4447
“ [=( 2 Jef 2 a]
Let A
o-(? Z )
2
vihere D is symmetric,
‘Therefore, we can write C=-[B"D+ DA]
Let us consider an autonomous system expressed as
For the system we can write
EHS ai}
Fad Leste ae,
Let C=(B'D+DB)=1
is (aye 2he aye {eRe
1 Mm, Dy Dy Dy Dy a My, 3 0
From the above equation, we can write
(19.43)
-2m,D,, =
(19.44)
Dy —™Di— Pig
2D,,-2m,D,, =-) (19.45)H
820 | control Systems
From Eqs. (19.43)-(19.45) we set
1
meemitm, me
ee
+1
1 =
Dye and Pa = 3mm,
sitive definite
To make D po: Pet
2 DyDn- Da?
From Eqs. (19.46) and (19.47) we get
mtmitm, 4
‘2mm,
and
nt +(m, +1] > 0 vine
ae oe
To make Eq. (19.49) true, m, > 0 and [{mz +(m, +1)?}/m}] is greater than zero.
(1945
(1947,
(19.48)
(19.49)
Here m,>0 and to make D positive definite, m, must be greater than zero. For stability fom
Lyapunov’s second method, m, > 0 and m, >0. In Lyapunov's first method we have already considered
m,>0 and m, must be greater than zero. Therefore, Lyapunov’s first method is highly conservative.
SIGNIFICANT POINTS
Non-Linear Control Systems: There are control systems which do not have any linearized version and its vey
0, = 1/2. Therefore, the describing function is given by
0 for|X|A
Off Non-Linearity with Hysteresis
=f eso
mX822 | Control Systems
Relay with Dead-Zone Hysteresis
4) (4) -(4) for |X|>4
4) +(2) eam ( By ett>
0 for|X|0- forall x#0
dF (x)
FO <9 forall x40
<0 forall:
‘Then, the system is asymptotically stable in the neighbourhood of the origin. On the other hand, if a = <0,
the system is stable but there is no assurance for asymptotic stability. The function F(x) is known as Lyapunov's
function.
Lyapunoy’s Second Method: If 4 is a positive matrix, we can write
P(x)=x7 Ar
BFC) tad era
or
Again i= Bx. Therefore 7 = x"BT,
Fe HT Att x" A= xBT Ax x ABr= x1 (BTA + AB)x
Let C=~(BTA+ AB)
Therefore, ee) =FQ)=-"'e
Selecting arbitrarily a positive definite or semi-definite matrix for C, we can solve for C. The necessary #0
sufficient condition for asymptotic stability is that the matrix Camust be positive definite,824 | Control systems
TEST YOURSELF
A non-linear control system
A=(0,0) and B=(x,0) will be, respectively,
(b) centre and focus
(@) centre and saddle
2. Alinear control system is shown below. Ithe intercept of Ga) on the negative real axis org, aa
lang
0) , the amplitude of the limit eycle is
(-
described by the equation 6+
sin @=0, The type of singly
(©) focus and saddle (qy Sada
Gs)
wok
ale
Figure 1
(@) m () 2m © 3m
3. For determination of stability of a non-linear control
G(j@) and -1/N(s) drawn as shown below for an input
(©) unstable and stable
P and Q will be respectively,
(@) stable and stable (b) stable and unstable
4. Non-linearities can be
(@) incidental
5. The intentional non-linear systems are
(@) On-off transducers
(©) non-linear pitch or dampers in airéraft control
6. Non-linearity occurs in servomotor due to
(a) dead space (b) static friction
(@) backlash (©) saturation
7. Introduction of non-linearities
(@) produces non-linear motion of the system
(b) intentional
(©) simplifies the system (4) all of theses
i
ANSWERS
1@20 3@ 4@ s@ 6@
(©) linearized
@ 4m
system by the deseribing function metho, te paso
x= X sin aw. The types of limit eyes atthe pone
(@ unstable and unstable
@) either (@) orb)
(b) contractors servos
@all of these
(©) coulomb friction
(b) achieves special characteristics,
(©) both (b) and (c)
7©| . a SECOND EDITION
Control Systems
mHEORY Aenee APPLICATIONS
TTY a
‘A Smarajit Ghosh
Fc) a” ae ee aT Oe
ee