This paper describes one method of research and synthesis of control systems for
the output of the object by the gradient-speed method of Lyapunov vector functions.
The task of the synthesis of the regulator and the observer is considered as a system
that can provide the specified (desired) transition characteristics of a closed system.
The gradient-speed method of Lyapunov vector functions allows one to solve the
problem of control systems directly using the elements of the matrix of the controller
object and the observer.
The approach to determine the range of changes in the object parameters, a
regulator, and an observer that provides the specified (desired) transient
characteristics of a closed system.
2. Research and Ssynthesis of Control Systems by Output of the Object by Speed - Gradient Method
of A.M.Lyapunov Vector Functions
http://www.iaeme.com/IJCIET/index.asp 121 editor@iaeme.com
combined with the observer’s own number [1,3,6]. This way synthesizing the modal
controller with the observer becomes a challenging task.
The known iterative algorithms [1,3] of separate own value control are based on the
preliminary matrix triangularization or block-diagonalization. Notice that the control matrix is
used as the nonsingular transition matrix in this canonical transformation, and it is defined by
its own vectors in complicated unequivocal ways described here [6,7].
The paper considers the control systems for the output of an object. To research the
dynamic equalizer we use the Lyapunov gradient-velocity vector function [8,9,10,11].
Construction of the vector function is based on the gradient nature of the control systems from
the theory of catastrophe [12,13]. Investigation of the closed-loop control system’s stability
and solution of the problem of controller (determining the coefficient of magnitude matrix)
and observer (calculation of the matrix elements of the observing equipment) synthesis is
based on the direct methods of Lyapunov [14,15,16]. The approach offered in this paper can
be considered as a way of determining the parameters of the controller and observer for a
closed-loop with certain transitional characteristics.
2. RESEARCH
In the present study tsunami forecast stations were selected for output of tsunami simulation
along the coast of India, Pakistan, Oman and Iran. Most of the tsunami forecast stations were
selected in such a way that sea depth is less than 10.0m to better
Assume the control system can be described by this set of equations [1,2,3,4]:
00 )(),()(),()()( xtxtCxtytButAxtx
, (1)
)(ˆ)( txKtu , (2)
00
ˆ)(ˆ),()()(ˆ)()( xtxtLytButxLCAtx
, (3)
Modify the state equation (1)-(3). For this we will use the estimation error )(ˆ)()( txtxt .
Then we can write is as: )()()(ˆ ttxtx , and equations (1)-(3) will transform to:
00 )(),()()()( xtxtBKtBKxtAxtx
, (4)
00 )(),(),()( ttLCtAt
(5)
consider the system with one input and one output, hence the system looks like:
nnnnn l
L
b
B
aaaa
A
0
...
0
0
,
0
...
0
0
,
...
0...000
...............
0...100
0...010
121
nn cccCkkkK ,...,,,,...,, 2121
The set of (4) , (5) will transform into:
3. М. А. Beisenbi, Zh. О. Basheyeva
http://www.iaeme.com/IJCIET/index.asp 122 editor@iaeme.com
nnnnnnnnnn
nn
nnnnnn
nnnnnnnnnn
nn
claclaclacla
kbkbkbkb
xkbaxkbaxkbaxkbax
xx
xx
xx
)(,...,)()()(
...
,...,
)(,...,)()()(
...
133222111
1
32
21
332211
133222111
1
32
21
(6)
Notice that in the absence of the external impact, the process in the set (4), (5) must
asymptotically approach the processes of a system with a controller, as if the closed-loop
system according to a state vector, was affected by the impact of the convergent disturbance
waves. These disturbances are caused by the )(tK polynom in the equation (5). The error
must converge and the speed of convergence is defined during the synthesis of the observer.
The main property of the set (4) and (5) lies in the asymptomatical stability. This way we
found the requirement for the asymptotical stability of the system using the gradient-velocity
method of the Lyapunov functions [8,9,10,11].
From (6) we find the components of the vector gradient for the Lyapunov vector function
:)),(),...,,(),,((),( 221 xVxVxVxV n
nnn
n
n
nn
n
nn
n
nn
nnn
n
n
n
n
n
n
nnn
n
n
nn
n
nn
n
n
n
n
cla
xV
cla
xV
cla
xV
xVxV
kb
xV
kb
xV
kb
xV
xkba
x
xV
xkba
x
xV
xkba
x
xV
x
x
xV
x
x
xV
x
x
xV
)(
),(
,...,)(
),(
,)(
),(
.
;...,
),(
;
),(
),(
,...,
),(
,
),(
,)(
),(
,...,)(
),(
,)(
),(
;
),(
;.
),(
;.
),(
1
2
221
2
2
11
1
2
3
2
2
2
1
1
22
2
11
1
1221
2
11
1
1
3
3
2
2
2
1
(7)
From (6) we find the decomposition of the velocity vector to the coordinates
).,...,,,...,( 11 nnxx
4. Research and Ssynthesis of Control Systems by Output of the Object by Speed - Gradient Method
of A.M.Lyapunov Vector Functions
http://www.iaeme.com/IJCIET/index.asp 123 editor@iaeme.com
,)(,...,)(,)(
...
;;;
,,...,,
,)(,...,)(,)(
,,...,,
122111
1
3
2
2
1
2211
122111
3
2
2
1
21
31
21
21
32
nnn
n
nn
n
nn
n
n
n
nnn
n
n
n
n
n
nnn
x
n
nn
x
n
nn
x
n
n
x
n
xx
cla
dt
d
cla
dt
d
cla
dt
d
dt
d
dt
d
dt
d
kb
dt
dx
kb
dt
dx
kb
dt
dx
xkba
dt
dx
xkba
dt
dx
xkba
dt
dx
x
dt
dx
x
dt
dx
x
dt
dx
n
n
n
n
n
(8)
To research the stability of the system (6) we use the basics of Lyapunov’s direct method
[14,15,16]. For the system to achieve the asymptotical equilibrium we need to secure the
existence of a positive function ),( xV so that its total derivative on the time axis along the
state function (6) is a negative function. The total derivative from Lyapunov function with
regard to the state equation (6) is defined as a scalar product of the gradient (7) from
Lyapunov and the velocity vector. (8):
22
1
2
2
2
21
2
1
22
3
2
2
222
2
2
2
2
22
1
2
1
222
1
2
3
2
32
2
2
2
21
2
1
2
1
22
3
2
2
2
1 11 1
)(,...,)()(,...,
)(,...,)()(
)(,...,
),(),(),(
nnnnnnnnnnn
nnnnnnnnn
nnn
n
ni
n
k
i
k
i
n
i
n
k x
i
k
i
claclaclakb
kbkbxkbaxkbaxkba
xkbaxxx
dt
dxV
dt
dx
x
xV
dt
xdV
kk
(9)
From (9) we derive that the total time derivative of the vector function will be negative.
Lyapunov function from (7) can be represented in the scalar view:
,)1(
2
1
,...,)1(
2
1
)1(
2
1
)(
2
1
)1(
2
1
,...,)1(
2
1
)1(
2
1
)(
2
1
)(
2
1
,...,)(
2
1
)(
2
1
)(
2
1
2
1
,...,
2
1
2
1
2
1
,...,
2
1
2
1
2
1
)(
2
1
,...,)(
2
1
)(
2
1
)(
2
1
2
1
,...,
2
1
2
1
),(
2
1
2
3332
2
2221
2
111
2
1
2
332
2
221
2
11
2
1
2
332
2
221
2
11
22
3
2
2
22
33
2
22
2
11
2
1
2
332
2
221
2
11
22
3
2
2
nnnnnnnnnnn
nnnnnnnnnn
nnnnnnnnnnn
nnnnnnnnnn
nnnnnnn
kbclakbclakbcla
kbclaxkbaxkbaxkba
xkbaсlaclaclacla
kbkbkbkbxkba
xkbaxkbaxkbaxxxxV
(10)
The condition for the positive certainty (10) i.e. existence of Lyapunov function will be
defined:
0>1
...
0>1
0>1
0>
1
32
21
1
nn
nn
nn
nn
kba
kba
kba
kba
(11)
0>1
...
0>1
0>1
0>
1
332
221
11
nnnn
nnn
nnn
nnn
kbcla
kbcla
kbcla
kbcla
(12)
5. М. А. Beisenbi, Zh. О. Basheyeva
http://www.iaeme.com/IJCIET/index.asp 124 editor@iaeme.com
The quality and the stability of the control system is dictated by the elements of the matrix
of the closed- system. That is determining the target values of coefficients in a closed-loop
system will prodivide smooth transitional processes in a system and result in higher quality
control.
The set of inequalities (11) and (12) serve as the necessary condition for the robust
dynamic equalizer. The condition (11) allows for the stability in the state vector.
Imagine a control system with a set of desired transition processes with one input and one
output:
nnnnn
nn
xdxdxdxdx
xx
xx
xx
132211
1
32
21
,...,
...
(13)
Explore the system (13) with the given coefficients n),...,1i( id , using the gradient-velocity
function [8].
From (13) we find the components of the gradient-vector function ))(),...,(()( 1 xVxVxV n
nn
n
n
n
n
n
n
n
n
n
n
n
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
x
x
xV
x
x
xV
x
x
xV
)(
,...,
)(
,
)(
,
)(
;
)(
;....,
)(
;.
)(
32
3
21
2
1
1
1
3
3
2
2
2
1
(14)
From (13) we determine the decomposition of the velocity vector according to the
coordinates:
n
x
n
n
x
n
n
x
n
n
x
n
n
x
n
xx
xd
dt
dx
xd
dt
dx
xd
dt
dx
xd
dt
dx
x
dt
dx
x
dt
dx
x
dt
dx
n
n
132211
1
3
2
2
1
,...,,,
;;....,;.
321
32 (15)
The total time derivative from the vector function is defined as a scalr product of the
gradient vector (14) and the velocity vector (15):
22
1
2
3
2
2
2
2
2
1
2
1
22
32 ,...,;....,
)(
nnnnn xdxdxdxdxxx
dt
xdV
(16)
From (16) follows that the total time derivative of the vector-gradient function is negative.
The vector function in the scalar view from (14) can be represented as:
2
1
2
32
2
21
2
1 )1(
2
1
,...,)1(
2
1
)1(
2
1
2
1
)( nnnn xdxdxdxdxV
(17)
Now the task is to define the controller coefficients (the elements of the matrix K) so that
they have values id .
This way exploring the stability of the system with set coefficients id will result in:
0,>1-0,...,>1-0,>1-0,> 121 dddd nnn (18)
Equalizing the elements of the set of inequalities (11) and (18) we will receive:
6. Research and Ssynthesis of Control Systems by Output of the Object by Speed - Gradient Method
of A.M.Lyapunov Vector Functions
http://www.iaeme.com/IJCIET/index.asp 125 editor@iaeme.com
)a-(
1
...
)a-(
1
)a-(
1
)a-(
1
11
2-n23
1-n12
n1
d
b
k
d
b
k
d
b
k
d
b
k
n
n
n
n
n
n
n
n
(19)
0>12
...
0>1d-c2
0>1d-c2
0>2
11
2-n3n2
1-n2n1
1
dcla
la
la
dcla
nn
n
n
nnn
(20)
From the set of inequalities (20) we will receive:
,1,...,2,1,
21d
max,
2d
max
1
i-n
1
n
n
ni
c
a
c
a
l
i
in
i
n (21)
This way the equations (4) and (5) allow us to conclude that in the absence of the external
impact, the process in the system asymptotically approaches those in the system with a
controller which state vector was affected by convergent disturbances. The role of the
disturbances is played by the compound )(tBk in the equation (4).
The speed of convergence of the error )(t can be determined during the synthesis of the
observer from (21).
A system with m-input and n-output is considered, respectively, the output control system of
the object has a matrix of the form
nnnnn
n
n
aaaa
aaaa
аааа
A
321
2232221
1131211
,
nmnnn
m
m
bbbb
bbbb
bbbb
B
321
2232221
1131211
,
ln321
2232221
1131211
...
...............
...
...
cccc
cccc
cccc
C
lll
n
n
,
)(
...
)(
)(
)( 2
1
tx
tx
tx
tx
n
,
)(
...
)(
)(
)( 2
1
t
t
t
t
n
,
mnmmm
n
n
kkkk
kkkk
kkkk
K
...
...............
...
...
321
2232221
1131211
, nlnnn
l
l
llll
llll
llll
L
...
...............
...
...
321
2232221
1131211
Us (4) and (5) can be expanded as
7. М. А. Beisenbi, Zh. О. Basheyeva
http://www.iaeme.com/IJCIET/index.asp 126 editor@iaeme.com
l
i
njnijn
l
j
l
j
jijjij
l
j
jiji
m
j
m
j
m
j
m
j
njnijijij
m
j
jijjijnjnijn
m
j
m
j
jijjij
m
j
jiji
nicla
claclacla
nikbkbkbkbxkba
xkbaxkbaxkbax
1
1
1 1
33132212
1
1111
1 1 1 1
3
1
22111
1 1
33132212
1
1111
,...,1,)(...
)()()(
;,...,1,)(...)()()()(...
)()()(
(22)
We find the condition of robust asiptotic stability of the system (22) by the gradient-
velocity method of Lyapunov vector functions [8,9].
From (22) we find the components of the gradient vector for Lyapunov vector functions.
nknicla
xV
nknikb
xV
nknixkba
xV
l
j
kjkijik
k
n
m
j
kjkij
k
i
m
j
kjkijik
k
i
,...,1;,...,1,)(
),(
,...,1;,...,1,)(
),(
,...,1;,...,1,)(
),(
1
1
1
1
(23)
(23) we decompose the velocity vector to coordinates ),...,,...,( 11 nnxx
nknicla
dt
d
nknikb
dt
dx
nknixkba
dt
dx
l
j
kjkijik
i
m
j
kjkij
i
m
j
kjkijikx
i
k
k
k
,...,1;,...,1,)()(
,...,1;,...,1,)()(
,...,1;,...,1,)()(
1
1
1
(24)
The total time derivative of Lyapunov functions is defined as the scalar product of the
gradient vector (23) and the velocity vector (24)
n
i
n
k
l
j
kjkijik
n
i
n
k
m
j
kjkij
n
i
n
k
m
j
kjkijik
n
i
n
k
i
k
i
n
i
n
k
i
k
i
n
i
n
k x
i
k
i
clakbxkba
dt
d
x
xV
dt
dx
x
xV
dt
dx
x
xV
dt
xdV
kkk
1 1 1
22
1 1 1
22
1 1 1
22
1 11 11 1
,)()()(
),(),(),(),(
(25)
From (25) it follows that the total time derivative of Lyapunov vector functions is a sign-
negative function. From (23), the Lyapunov function can be represented as:
,)])[(
2
1
)(
2
1
),( 2
1 1 111 1 1
2
k
n
i
n
k
m
i
jkij
l
j
ikjkij
n
i
n
k
m
j
kikjkij kbaclxakbxV
(26)
The condition for the existence of a Lyapunov function is defined by the inequalities:
nkniakb
n
j
ikjkij ,...,1;,...,1,0>
1
(27)
8. Research and Ssynthesis of Control Systems by Output of the Object by Speed - Gradient Method
of A.M.Lyapunov Vector Functions
http://www.iaeme.com/IJCIET/index.asp 127 editor@iaeme.com
nkniakbacl
m
j
ikjkij
n
j
ikjkij ,...,1;,...,1,0>-
11
(28)
The system of inequalities (27) and (28) is a condition for the robust stability of the
dynamic compensator. Condition (27) allows for robust stability of the system when
controlled by a state vector. Suppose we have a closed control system of desired transients
with m-inputs and n-outputs with matrix:
nnnnn
n
n
dddd
dddd
dddd
321
2232221
1131211
G
The task is to determine the coefficient of the regulator (elements of the matrix K) such
that the elements of the matrix of a closed system have the specified values.
We study the stability of a system with given values of the coefficients by the gradient-
speed method of Lyapunov vector functions.
For a system with given parameters, write the equations of state in expanded form:
nnnnnnn
nn
nn
xdxdxdxdx
xdxdxdxdx
xdxdxdxdx
,...,
...
,...,
,...,
332211
23232221212
13132121111
(29)
From (29) we find the components of the gradient vector of Lyapunov vector functions
))(),...,(()( 1 xVxVxV n
nnn
n
n
n
n
n
n
n
n
nn
n
nn
n
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
xd
x
xV
)(
,...,
)(
,
)(
,
)(
...
)(
,...,
)(
,
)(
,
)(
)(
,...,
)(
,
)(
,
)(
33
3
22
2
11
1
2
2
323
3
2
222
2
2
121
1
2
1
1
313
3
1
212
2
1
111
1
1
(30)
The total derivative of the Lyapunov vector-function with respect to time is determined by
the expression:
222
3
2
3
2
2
2
2
2
1
2
1
22
2
2
3
2
23
2
2
2
22
2
1
2
21
22
1
2
3
2
13
2
2
2
12
2
1
2
11
,...,
,...,,...,,...,
)(
nnnnnn
nnnn
xdxdxdxd
xdxdxdxdxdxdxdxd
t
xV
(31)
From (31) it follows that the total time derivative of Lyapunov vector functions is a sign-
negative function.
From (30), the Lyapunov function can be represented in scalar form:
9. М. А. Beisenbi, Zh. О. Basheyeva
http://www.iaeme.com/IJCIET/index.asp 128 editor@iaeme.com
2
21
2
222212
2
112111
),...,(
2
1
,...,),...,(
2
1
),...,(
2
1
)(
nnnnn
nn
xddd
xdddxdddxV
(32)
The condition for the existence of Lyapunov vector functions for system (29) is described
by the system of inequalities:
0,...,
...
0,...,
0,...,
0,...,
21
32313
22212
12111
nnnn
n
n
n
ddd
ddd
ddd
ddd
(33)
We can equate inequalities (22) for system (27) and (33) to have the specified properties
and from there find the required values of the coefficients of the matrix k:
nknidakb ik
n
j
ikjkij ,...,1;,...,1,
1
(34)
nknidakb ik
n
j
ikjkij ,...,1;,...,1,0
1
(35)
From this n-system of algebraic equations (35) we can find the values of the elements of
the matrix k ),...,1;,...,1;( njnikij
.
System (28) can be rewritten in the form:
nknidaсl ik
n
j
ikjkij ,...,1;,...,1,02
1
(36)
Solving the system (36) with respect to the coefficients ),...,1;,...,1; linjlij
, we can find the
boundary values for the coefficients of the observing device.
Suppose, for simplicity, we assume that the matrices B, L and C have dimension,
respectively 1,1,1 lln , and the matrices A and D are given a controlled canonical form.
Then the equations can be rewritten in the form:
;,...,1,0)(
1
1
nida
b
k i
n
j
i
i
j
(37)
From (37), the boundary values of the coefficients ),...,1;( lili
of the observability matrix
will be determined:
ni
k
da
l n
j
j
ii
i ,...,1,
2
1
(38)
10. Research and Ssynthesis of Control Systems by Output of the Object by Speed - Gradient Method
of A.M.Lyapunov Vector Functions
http://www.iaeme.com/IJCIET/index.asp 129 editor@iaeme.com
Thus, the problem of ontrol for the output of the object is solved by the gradient-speed
method of Lyapunov vector functions completely. The calculation of the coefficients of the
controller and the observing device according to the desired characteristic of the system (29)
is presented as a solution of the system of algebraic equations (35) and (36).
The calculation of the matrix elements of the regulator and the observer requires certain
regulated transformations - complex and ambiguous calculations of the roots of the
characteristic equation of a closed system. The roots of the characteristic polynomial of a
closed system are obtained by combining the roots of the system with a model controller and
the eigenvalues of the state observer.
3. CONCLUSION
The known methods of synthesis of the systems with controllers that use the state value and
the observer are based on the combination of the roots of the characteristic polynomial with a
modal control with the observer’s own number. This process requires substantial calculations
and requires preliminary matrix block-diagonalization. This nonsingular matrix in the
canonical transformation is defined by its own vectors in.
Researching the closed-loop control system using the gradient-velocity method of
Lyapunov function gave us the opportunity to develop an approach for controller and observer
parametrization that provides us with a system of our desire without extraneous calculations.
REFERENCES
[1] Andrievsky BR, Fratkov AL Selected chapters of the theory of automatic control with
examples in the language MATLAB - SPb .: Nauka, 2000. 475p.
[2] Kvakernah H., Sivan R. Linear optimal control systems. M .: Mir, 1986. 650 p.
[3] Andreev Yu.N. Managing finite-dimensional linear objects. M .: Nauka, 1976. 424 p.
[4] Rey U. Techniques for managing technological processes. M .: Mir, 1983.638 p.
[5] Kukharenko N.V. Synthesis of modal regulators with incomplete controllability of objects
// Izvestiya Akademii Nauk. Russian Academy of Sciences. Technical cybernetics. No. 3
[6] Gantmakher FR Theory of matrices. Moscow: Nauka, 1967.
[7] Streitz, W., The method of the space of states in the theory of discrete linear control
systems, Per. with English. Moscow: Nauka, 1985.
[8] Beisenbi M.A. Investigation of robust stability of automatic control systems by the
method of functions of AM Lyapunov .- Astana, 2015. 204 p.
[9] Beisenbi M., Uskenbayeva G. The New Approach of Design Robust Stability for Linear
Control System. Proc. of the Intl. Conf. on Advances in Electronics and Electrical
Technology—AEET 2014, pp.11-18. 2014
[10] Beisenbi M., Yermekbayeva J. Construction of Lyapunov function to examine Robust
Stability for Linear System. International Journal of Control. Energy and Electrical
Engineering (CEEE), V (1) pp.17-22. Publisher Copyright – IPCO-2014.
11. М. А. Beisenbi, Zh. О. Basheyeva
http://www.iaeme.com/IJCIET/index.asp 130 editor@iaeme.com
[11] Beisenbi M., Uskenbayeva G., Satybaldina D., Martsenyuk V., Shailhanova A. Robust
stability of spacecraft traffic control system using Lyapunov functions. 16th International
Conference on Control, Automation and System (ICCAS), IEEE 2016, pp. 743-748.
[12] Gilmore R. Applied theory of catastrophes. In 2 volumes. T.1. - Moscow: The World,
1984.
[13] Poston T., Stuart I. The theory of catastrophes and its applications. - Moscow: Nauka,
2001, No. 6
[14] Malkin I.G. Theory of stability of motion. - Moscow: Nauka, 1966. 534 p.
[15] EA Barbashin, "Introduction to the theory of stability", Nauka, Moscow, 1967. 225 p.
[16] Voronov AA, Matrosov V.М. The method of Lyapunov vector functions in stability
theory - M.: Nauka,1987. 252 p.