Вы находитесь на странице: 1из 3

Application of Fractional-order Calculus Approach to

Signal Processing

Zheng, Wang Xianmin, Ma


School of Electric and Control Engineering School of Electric and Control Engineering
Xi’an University of Science and Technology Xi’an University of Science and Technology
Xi’an, China Xi’an, China
e-mail:wendy7830@126.com e-mail:Maxm@xust.edu.cn

Abstract— As a new research field on controlling science and


engineering, fractional-order control gets more and more II. FRACTIONAL-ORDER CALCULUS
concerning. This article provides an overview of fractional-order
control of background and basic knowledge of mathematics, A. Mathematical operator
giving different definitions and numerical algorithms of FOC operator can be expressed as:
fractional calculus, finally establishing a mathematical model of
fractional-order systems which used for analyzing and ­ dD , R (D ) ! 0
researching image signal processing of fractional-order systems. °° dt D
Keywords- Fractional Calculus, Signal Processing, fractional- a DtD ® 1, R (D ) 0 (1)
° I D
³ (dW ) .R(D )  0
order Control(FOC) D

¯°
I. INTRODUCTION Where a and t are operating superior and lower limits,
Fractional-order calculus is about any first-order theory of
respectively. D is orders, R(D ) is real part of D .
differential and integral, which is unified with integer-order
calculus and is an generalization of integer-order calculus.
Many years ago it was created, but in recent years it is widely B. Fourier transform domain definition
used in the disciplines of different areas. At present, in According to the nature of the Fourier transform, v -order
electrochemistry, thermodynamics, mechanical engineering,
is productivity operator of v -
v
acoustics, and electromagnetic field and so on in different differential operator Dv D
areas, we are researching to establish a mathematical models
of system by fractional differential equation, so that can order Differential multiplier function dˆ (Z ) (iZ ) v , the
describe the system's dynamic characteristic better. These exponential form of fractional calculus in a complex field is:
models are typically fractional differential equations.
­dˆ (Z ) (iZ ) v aˆ v (Z ) x exp(iT v (Z ))
With the development of modern technology, especially in °°
®a(Z ) x p v (Z )
computer applications, FOC theory offers new theoretical ˆ ˆ (2)
bases and mathematical tools for many disciplines. At the
° aˆ (Z ) Z , T (Z )
v vS
same time, some necessary basic theories in engineering are °¯ v v sgn(Z )
also developed and researched appropriately, such as FOC 2
differentiability, calculation rules, numerical algorithms, and From the perspective of communication modulation, the
variation problems. In recent years, a number of scholars have physical meaning of signal fractional differential can be
been raised research interests to applied FOC to the control understood as a generalized phase-amplitude modulation, and
field. With the improvement continuously of FOC numerical its amplitude changes with frequency by fractional-order
methods and algorithms, and suggesting of various fractional- power-exponent, and the phase is generalized Hilbert
order analysis and control strategies, it is more impetuses in transform of the frequency. From the perspective of signal
the application of fractional-order control theories and rapid processing, v -order calculus operation actually is linear time-
development. invariant filter system to signal, its filter function is:
Fractional calculus has become a important branch of pure v
mathematics system. However, it is also rare that fractional dˆ v (Z ) (iZ ) v Z x exp(iT v (Z ))
calculus is applied in modern signal analyzing and processing
at home and abroad. C. Numerical algorithm based on G-L definition
Defined by G-L, with combinatorial numbers and the
nature of the Gamma function, approximation expression of

___________________________________
978-1-4244-8625-0/11/$26.00 ©2011 IEEE


derivatives of fractional-order that is defined by G-L as m
s ( k ) (a )(t  a )  v  k
follow: G
a
v
D s (t )
t ¦ *(v  k  1)

a ' h s (t )
v
~ 1 n v k 0
G
a D s (t )| Dtv
t
v G
a
hv
¦ wk s(t  kh) (3)
hk1 1
t
(6)

*(v  m  1) ³a
G ~ (t  W ) v  m s ( m 1) (W )dW
It has been certified that a Dtv converge in
G
a Dtv s (t ) With equivalent order infinitesimal precision Where Gamma functions
O(h) DŽ For first-order approximation of fractional f
*(D ) ³e x D 1 dx
x
v
derivative defined by G-L, weight coefficient wk happens to
(D  1)!
0 DŽ
be equal to the Taylor expansion coefficient of the power
series at the origin. Then v -order differential of c
The Fourier transformation of first-order derivative s (t )
Signal s (t ) is available:
of Any square integral Energy signals s (t )  L ( R )
2

d v
ntv v
*(k  v)
n 1
kt
s ( v ) (t )
dt v
¦
*(v) k 0 *(k  1)
s (t  )
n
(4) is Ds (t )
FT
œ ( Dsˆ)( w) (iw) x sˆ( w) d ( wˆ ) s ( wˆ ) .

Similarly, its Fourier transformation of v-order fractional


ªt  a º differential is:
Where n
« h » is the length of signal data.
¬ ¼
d v s (t ) FT
III. IMAGE SIGNAL PROCESSING ANALYSIS OF D v s (t ) Dv s (t ) œ ( Dv sˆ)( w)
FRACTIONAL-ORDER DIFFERENTIAL dt v (7)
CHARACTERISTIC (iw) v x sˆ( w) d v ( wˆ ) s ( wˆ ), v  R 
B Study on G-L of fractional calculus come from the v
classic definition of continuous function of integer order Where v-order differential operator Dv D is
derivative, of which order number and dimensions are productivity operator of v-order differential multiplier
extended from integer to fraction. v  R (including function dˆ ( w) (iw) v , and its complex exponential form
fraction), and its integer part v>@ ,if and time domain form at respectively are:
s (t )  >a, t @(a ¢t , a  R, t  R) has m  1 - order
signal ­dˆ (Z ) (iZ ) v aˆ v (Z ) x exp(iT v (Z ))
(m䌜 Z, Z is integer) continues derivative; when v ! 0 ,m °°
is >v @ at least, v -order derivative is:
®a (Z ) x p v (Z )
ˆ ˆ (8)
° aˆ (Z ) Z v ,Tˆ (Z ) vS sgn(Z )
' ' n
ª v º °¯ v v
2
G
a Dtv s (t ) lim s h( v ) (t ) lim h v ¦ « »s (t  rh)
h o0 h o0
nh t  a r 0¬ r ¼ f
(5) 1
³ (iZ ) x e iZt dZ (9)
v
d v (t ) a v (t ) * p v (t )
ª v º (v)(v  1) " (v  r  1) 2S f
Where « » DŽ If g is
f f
¬r ¼ r! 1
³ aˆ (Z ) x e dZ ³Z
i Zt v
extended in the combinatorial a v (t ) x cos(iZt )dZ (10)
S
ª g º
( g )( g  1) " ( g  r  1) f 0
number « » to any real f
¬ r ¼ r! 1
ª g º ªg º
p v (t )
2S ³ pˆ
f
v (Z ) x e iZt dZ
number (including fractions), then « » (1) r « » DŽTo (11)
¬ r ¼ ¬r ¼ vS vS 1
(v) cos x G (t )  sin x
s (t ) to reach non-zero limit, when hĺQĺ’ ,
enable the h 2 2 St
ta ªt  a º According to formula (8), from the perspective of
so assumed that h ,then n
« h » DŽ Firstly communication modulation, the physical meaning of signal
n ¬ ¼ fractional calculus can be understood as a generalized phase-
mathematical induction of (5), and then partial integration is amplitude modulation, and its amplitude fractional power
available: variation with the frequency, and its phase is generalized
Hilbert transformation of frequency. According to (9)-(11),


filter function for fractional differential filter
Z x exp>iT (Z )@
v
dˆ v (Z ) (iZ ) v
v
is , and whose IV. CONCLUSION
amplitude characteristic is an even function and phase
In the field of information science, especially in research
characteristic is an odd function. So only when Z!0 , of modern signal processing at home and abroad, fractional
d v (t ) filter characteristic can be researched. calculus are also emerging. Application of Fractional calculus
in signal processing, especially in digital image processing , is
According to Figure 1, from the perspective of signal a new discipline study branch. This article considers only
processing, v-order differential operator is a linear time- about some preliminary attempts about signal analysis and
invariant filter of signal. In the time domain, convolution of processing of fast numerical realization of fractional calculus,
many problems remain to be further studied.
fractional calculus of signal s (t ) is

Dv s(t ) d v (t ) s(t ) a v (t ) p v (t ) s (t ) REFERENCES


f (15)
ª vS vS 1 º
³ a v (t  W )s(W )dW
[1] Podlubny I. Numerical solution of ordinary fractional differential
«cos x G (t )  sin x »
¬ 2 2 St ¼ f
equations by the fractional difference method [ C ] . Proc of the 2 # Int
Conf in Difference Equations. London ,1997 : 5072516.
[2] Zhang S Q. The existence of positive solution for a nonlinear fractional
differential equation [ J ] . Math Analysis Applied , 2000 , 252 (4) :
8042812.
[3] JBabakhani A , Daf tardar2Gejji V. Existence of positive solutions of
nonlinear fractional differential equation [J ] . Math Analysis Applied ,
2003 , 278 (2) : 4342442
[4] Planka L, Mrozek A.Rule-based stabilization of the inverted
pendulum[J]. Cpmputational Intelligence, 2005,11(2):348~356.
[5] Czogala E.Idea of a rough fuzzy controller and its application to the
stabilization of a pendulum-car system[J]. Fuzzy Sets and
Systems,2005,72(1):61~73.
[6] Takagi T, Sugeno M. Fuzzy identification of systems and its
applications to modeling and control [J].IEEE trans Syst,
Man,Cybern,2006, 15:116~132.

Figure 1. Filter function of fractional differential

So when V=0 it is not differential nor integral but an all-

pass filter,
dˆ v (Z ) { 1 œ d v (t ) G (t ) ; when v  0 ,it is an

integrator,
dˆ v (Z ) is a singular low-pass filter; when v ! 0 ,it

is differential operation to signal , dˆ v (Z ) Is a


lim dˆv (Z ) of
Z of

singular high-pass filter. The more v is large, the more pass


band is narrow and the high-pass characteristic is obvious.
High frequency components of signal s (t ) are composed and
its low frequency components are suppressed relatively, which
is conducive to highlight details of the signal, but composition
performance of anti-high-frequency interference will be
become worse.



Вам также может понравиться