3 views

Uploaded by salim

FO systems

- CL7103_SystemTheoryquestionbank
- Li&Kumar2011revised
- CL7103-System Theory QB
- Control of Quadrotor Using Sliding Mode Disturbance Observer and Nonlinear H-Infinite
- A Mechatronic Analysis and Synthesis of Human Walking Gait
- Adaptive Integral-type Sliding Mode Control for Spacecraft Attitude
- ABS Modelling Ref Paper (1)
- 1-s2.0-S0005109806004432-main.pdf
- 125_600
- 06Nov2012121035Sliding Mode Controllers....
- detectability subspaces
- Springer Atif.sanjeev 2017
- Mozell i 2009
- Stochastic Stability Properties
- Adaptive control
- Industrial View on Control Teaching Theory
- Advanced PID Control_AstromHagglund_Chapter1 Introduction
- el07-106
- PDF v f Using PID and Encoder En
- SCE-Advanced PID Control S7-1200 (2016)

You are on page 1of 14

DOI 10.1007/s11071-015-2136-8

ORIGINAL PAPER

design for 3D fractional-order nonlinear systems

Chun Yin · Yuhua Cheng · YangQuan Chen ·

Brandon Stark · Shouming Zhong

© Springer Science+Business Media Dordrecht 2015

Abstract In this paper, an adaptive sliding mode tech- posed control scheme are illustrated through numerical

nique based on a fractional-order (FO) switching-type simulations.

control law is designed to guarantee robust stability

for uncertain 3D FO nonlinear systems. A novel FO Keywords Fractional-order switching-type control

switching-type control law is proposed to ensure the law · Sliding mode control · Reaching time ·

existence of the sliding motion in finite time. Appro- 3D fractional-order nonlinear system ·

priate adaptive laws are shown to tackle the uncertainty Adaptive sliding mode technique

and external disturbance. The calculation formula of

the reaching time is analyzed and computed. The reach-

ability analysis is visualized to show how to obtain a 1 Introduction

shorter reaching time. A stability criterion of the FO

sliding mode dynamics is derived based on indirect Fractional calculus [1] has gained a lot of attention in

approach to Lyapunov stability. Advantages of the pro- the past few years. The reason is that the mathematical

technology for FO operators has rapidly grown [2–7]

and FO modeling has been wildly used in real-world

This work was supported by National Basic Research Program applications [8,9]. In particular, fractional calculus has

of China (Nos. 61462065 and 51407024) and

ZYGX2015KYQD020.

also played a significant role in control theory. The main

advantage of FO controllers is the greater flexility in

C. Yin (B) · Y. Cheng improving the robustness and control performance. It

School of Automation Engineering, University of has motivated renewed interest in various fields of FO

Electronic Science and Technology of China, Chengdu,

control [10–14].

Sichuan, People’s Republic of China

e-mail: yinchun.86416@163.com Sliding mode control (SMC) approach is a robust

nonlinear control law that forces the states of the sys-

Y. Chen · B. Stark tem from the initial states onto some predefined switch-

Mechatronics, Embedded Systems and Automation

(MESA) Lab, School of Engineering, University of

ing surface [15,16]. The SMC methodology has been

California, Merced, 5200 North Lake Road, Merced, designed for FO chaotic systems in many published

CA 95343, USA results [17–26]. Some applications of FO SMC have

been investigated [27–29]. Yin et al. [22] investigated

S. Zhong

School of Mathematics Science, University of Electronic Science

the FO SMC for a general FO nonlinear system which

and Technology of China, Chengdu 611731, People’s Republic can represent many FO chaotic systems. Aghababa

of China et al. [26] utilized the terminal FO SMC to reach sta-

123

C. Yin et al.

bilization of FO systems. There are limited results in which n − 1 ≤ β̃ < n where n is an integer and

reported on improving the control performance of FO Γ (·) is the Gamma function. Likewise, the β̃th-order

SMC of FO systems. It should be mentioned that one integral of the function ζ (t) is given by

important aspect of a good SMC (including FO SMC) t

1

is to obtain a faster and lower chattering performance. D −β̃ ζ (t) = I β̃ ζ (t) = (t − τ )β̃−1 ζ (τ )dτ ,

Γ (β̃) 0

Hence, it is important to investigate how to design sys-

(2)

tematic methods to derive a better performance of FO

SMC. Research in this area should be interesting and in which ζ (t) denotes any integrable function and I β̃

challenging. represents the β̃th-order fractional integral on [0, t].

With this motivation, an FO switching-type con-

Definition 2.2 [30] Let (t) be the impulse response

trol law is designed for sliding mode technique of

of a linear system. If μ(ω) denotes the diffusive rep-

uncertain 3D FO nonlinear systems with the exter-

resentation (or frequency weighting function) of (t),

nal disturbance. For reachability analysis, a concept

(t) and μ(ω) satisfy the pseudo-Laplace transform

of FO sign function D β̄ sgn(·), 0 ≤ β̄ < 1, involv-

definition

ing an FO differentiator, is chosen to build an FO

∞

switching-type control law. It can ensure the occur-

rence of the reaching phase in finite time. The cal- (t) = μ(ω)e−ωt dω. (3)

0

culation formula of the reaching time treach under

FO switching-type control law is obtained. The rela- Remark 2.1 [30] Considering the FO integral I β̃ ζ (t),

tionship between treach , β̄ and the other parameters Eq. (3) can be expressed by I β̃ ζ (t) = (t) ∗ ζ (t),

is analyzed to reveal the potential advantages of FO

in which (t) = t β̃−1 /Γ (β̃) and ∗ denotes convolu-

switching-type control law. For the stability analysis,

tion operator. The diffusive representation of (t) =

an FO switching surface is designed for FO nonlinear

t β̃−1 /Γ (β̃) is given by μ(ω) = (sin(β̃π )/π )ω−β̃ .

systems. By applying continuous frequency distribu-

tion theory, stability analysis of the 3D FO nonlinear Definition 2.3 [30] Consider the nonlinear fractional

sliding mode dynamics is obtained based on indirect differential equation D β̃ η = ζ (η). From the continu-

approach to Lyapunov stability. Simulation results are ous frequency-distributed plant of the fractional inte-

provided to show the effectiveness of the designed con- grator, the above equation can be rewritten as

trol scheme. ∂ρ(ω, t)

The paper is presented as follows: In Sect. 2, basic = −ωρ(ω, t) + ζ (η(t)),

∂t

definitions and notations are given. In Sect. 3, the gen- ∞

eral form of 3D FO nonlinear systems is presented. η(t) = μ(ω)ρ(ω, t)dω, (4)

0

Section 4 proposes the sliding mode control method

with FO switching-type control law, including the sta- in which μ(ω) is defined in Remark 2.1.

bility analysis and the reachability analysis. Simula- Lemma 2.1 Consider D β̄ σ (t)= Γ (1−

1 d t σ (τ )

dτ

β̄) dt 0 (t−τ )β̄

tion results are shown in Sect. 5. Finally, conclusion is

with the fractional-order 0 ≤ β̄ < 1 and the sign func-

addressed in Sect. 6.

tion sgn(·), one can conclude that

β̄ > 0, if σ (t) > 0, when t > 0,

D sgn(σ (t))

2 Basic definition and preliminaries < 0, if σ (t) < 0, when t > 0.

Proof See “Appendix”.

To discuss uncertain 3D FO nonlinear systems, some

important definitions and lemmas are presented. Remark 2.2 The concept of D β̄ sgn(σ ),

involving an

FO differentiator, is proposed to extract the sign of σ .

Definition 2.1 [1,2] The Riemann–Liouville defini-

The sign of D β̄ sgn(σ ), 0 ≤ β̄ < 1 is firstly proven

tion of the β̃th-order derivative is

n t to be the same as sgn(σ ) in our previous work [14].

1 d ζ (τ ) Then, more investigations about reachability analysis

D β̃ ζ (t) = dτ ,

Γ (n − β̃) dt 0 (t − τ )1+β̃−n based on D β̄ sgn(σ ), including the calculation formula

(1) of reaching time, will be shown.

123

Adaptive fractional-order switching-type control method design

derived

A class of three-dimensional FO nonlinear systems is

considered as follows u eq (t) = −g(x, y, z)−x· f (x, y, z)−z·h(x, y, z)−ky.

⎧ q (8)

⎨ D 1 x = y · f (x, y, z) + z · ϑ(x, y, z) − αx,

D q2 y = g(x, y, z) − βy, (5) Furthermore, the following FO sliding mode dynam-

⎩ q3

D z = y · h(x, y, z) − x · ϑ(x, y, z) − γ z, ics can be obtained

⎧ q

where qi (i = 1, 2, 3) denote fractional orders satisfy- ⎨ D 1 x = y · f (x, y, z) + z · ϑ(x, y, z) − αx,

ing 0 < qi < 1. x, y and z are state variables. α, β D q2 y = −x · f (x, y, z) − z · h(x, y, z) − βy − ky,

⎩ q3

and γ are nonnegative known constants. f (·), g(·) and D z = y · h(x, y, z) − x · ϑ(x, y, z) − γ z.

h(·), ϑ(·) are continuation nonlinear vector functions, (9)

which belong to R 3 → R space.

Next, based on the FO sign function D β̄ sgn(s), a

Remark 3.1 Note that many FO chaotic systems can be novel FO switching-type control law is specified to

characterized by (5). Examples include that FO finan- ensure the existence of the sliding motion

cial system and the unified chaotic system of FO ver-

sion (including the FO Chen system, FO Lorenz sys- u sw = −λ|s|ι D β̄ (sgn(s)), (10)

tem). Table 1 shows that these FO chaotic models can

be described by (5). in which 0 ≤ β̄ < 1, 1 > ι > 0 and λ > 0 will be

chosen later. Then, based on (8) and (10), the overall

To control chaos of (5), the controller u(t) is added to FO sliding mode control signal is presented by

the second state equation. The FO system (5) is rewrit-

u = −g(x, y, z) − x · f (x, y, z) − z · h(x, y, z)

ten as

− ky − λ|s|ι D β̄ (sgn(s)). (11)

⎧ q

⎨ D 1 x = y · f (x, y, z) + z · ϑ(x, y, z) − αx, In the following, the stability analysis and reachabil-

D q2 y = g(x, y, z) − βy + u(t), (6) ity analysis will be discussed. First of all, the stability

⎩ q3

D z = y · h(x, y, z) − x · ϑ(x, y, z) − γ z, of the FO sliding mode dynamics is considered.

It is assumed that f (·), g(·), h(·) and ϑ(·) can guar- Theorem 4.1 The FO sliding mode dynamics described

antee that the FO system (5) with u(t) has a unique in (9) is globally asymptotically stable for arbitrary

solution in [T, +∞], the initial time T > 0 for any k > 0.

given initial condition.

Proof According to the frequency-distributed model of

fractional integral in Definition 2.3, the FO system (9)

can be rewritten as

4 Sliding mode technique

∂ρ1 (ω, t)

= −ωρ1 (ω, t) + y f (x, y, z) + zϑ(x, y, z)

Firstly, an FO switching surface s(t) is chosen as ∂t

− αx,

∞

s(t) = D q2 −1 y(t) + D −1 ϕ(t)

t x(t) = μ1 (ω)ρ1 (ω, t)dω, (12)

q2 −1 0

=D y(t) + ϕ(τ )dτ , (7) ∂ρ2 (ω, t)

0 = −ωρ2 (ω, t) − x f (x, y, z) − zh(x, y, z)

in which ϕ(t) is a function described by ϕ(t) = x · ∂t

− (β + k)y,

f (x, y, z) + z · h(x, y, z) + βy + ky, with k > 0. ∞

When the system (6) achieves to the switching surface, y(t) = μ2 (ω)ρ2 (ω, t)dω, (13)

one has 0

∂ρ3 (ω, t)

s(t) = D q2 −1 y(t) + D −1 ϕ(t) = 0, = −ωρ3 (ω, t) + yh(x, y, z)

∂t

ṡ(t) = D q2 y(t) + ϕ(t) = 0. − xϑ(x, y, z) − γ z,

123

C. Yin et al.

Table 1 List of published FO nonlinear chaotic systems, which can be described by the above model (5)

Name Mode f (x, y, z) g(x, y, z) h(x, y, z) ϑ(x, y, z)

⎧ q

⎨ D 1 x = a(y − x),

Chen’s system D q2 y = dx − x z + cy, a dx − x z x 0

⎩ q3

D z = x y − bz,

⎧ q

⎨ D 1 x = a(y − x),

Lorenz model D q2 y = x(b − z) − y, a x(b − z) x 0

⎩ q3

D z = x y − cz,

⎧ q

⎨ D 1 x = z + (y − a)x,

Financial system D q2 y = 1 − by − x 2 , x 1 − x2 0 1

⎩ q3

D z = −x − cz,

⎧ q

⎨ D 1 x = a(y − x),

Lu’s model D q2 y = −x z + cy, a −x z x 0

⎩ q3

D z = x y − bz,

⎧ q

⎨ D 1 x = −ax − ey 2 ,

Liu system D q2 y = −kx z + by, −ey −kx z mx 0

⎩ q3

D z = mx y − cz,

⎧ q

⎨ D 1 x = −ax + by − yz,

Rucklidge oscillator D q2 y = x, b−z x y 0

⎩ q3

D z = −z + y 2 ,

⎧ q

⎨ D 1 x = z,

Rossler Founder system D q2 y = ax − ax 2 − by, 0 ax − ax 2 −1 1

⎩ q3

D z = −y − x,

∞

z(t) = μ3 (ω)ρ3 (ω, t)dω, (14) Substituting (12) into (17), one has

0 ∞

dV1 (t)

in which μi (ω) = ((sin(qi π ))/π )ω−qi , i

= 1, 2, 3. = μ1 (ω)ρ1 (y f + zϑ − αx)dω

dt 0

Then, define two classes of Lyapunov functions: ∞

− ωμ1 (ω)ρ12 dω

(1) υi (ω, t)(i = 1, 2, 3) are the monochromatic Lya- 0

∞

punov functions corresponding to the elementary

= (y f + zϑ − αx) μ1 (ω)ρ1 dω

frequency ω; 0

∞

(2) Vi (t)(i = 1, 2, 3) are the Lyapunov functions

summing all the monochromatic υi (ω, t) with the − ωμ1 (ω)ρ12 dω (18)

0

weighting function μi (ω). The details are listed as

follows Then, one can derive

ρi2 (ω, t) ∞

dV1 (t)

υi (ω, t) = , (15) =− ωμ1 (ω)ρ12 dω + x(y f + zϑ − αx).

∞ 2 dt 0

Vi (t) = μi (ω)υi (ω, t)dω, i = 1, 2, 3. (16) (19)

0

Similarly, one has

Thus, the time derivative of the above Lyapunov func-

tions can be obtained ∞

dV2 (t)

=− ωμ2 (ω)ρ22 dω

∂υi (ω, t) ∂υi (ω, t) ∂ρi ∂ρi dt

= = ρi , 0

∂t ∂ρi ∂t ∂t + y(−x f − zh − (β + k)y),

∞ ∞

dVi (t) ∂υi (ω, t) dV3 (t)

= μi (ω) dω, i = 1, 2, 3. =− ωμ3 (ω)ρ32 dω + z(yh − xϑ − γ z).

dt 0 ∂t dt 0

(17) (20)

123

Adaptive fractional-order switching-type control method design

Defining V (t) = V1 (t) + V2 (t) + V3 (t), one can − λ|s|ι D β̄ (sgn(s)) + x · f (x, y, z)

conclude + z · h(x, y, z)]

3 ∞

dV (t) = − 2λs|s|ι D β̄ (sgn(s)). (23)

=− μi (ω)ωρi dω

2

dt 0 According to Lemma 2.1, kV̇ = 2s(t)ṡ(t) < 0. Thus,

i=1

−(αx + (β + k)y + γ z ).

2 2 2

(21) the states trajectories under the sliding mode tech-

3 3 ∞ nique with the FO switching-type control law can reach

Let Ψ = i=1 Ψ i = −

∞ i=1 ( 0 μi (ω)ωρi dω),

2

the predefined switching surface. This completed our

in which Ψi = − 0 μi (ω)ωρi dω, (i = 1, 2, 3).

2

proof.

Hence, the frequency discretizations of Ψi can be

obtained as follows Finite reaching time is one of the main characteris-

tics of SMC. The calculation formula of reaching time

J

will be explained later.

Ψi = − ωi j μi (ωi j )ρi2 (ωi j , t)Δωi j

j=1 Remark 4.1 From (23), we have 0.5V̇ = −λs|s|ι D β̄

= ωi j li j ρi2 (ωi j , t), sider two cases:

j=1 (1) When the initial condition s(0) is bigger than zero

where li j = −μi (ωi j )Δωi j < 0. (s(0) > 0),

It is obvious that Ψi (i = 1, 2, 3) are negative def-

inite quadratic forms and can be written as Ψi = ṡ = −λs ι D β̄ sgn(s). (24)

ΛiT L i Λi , i = 1, 2, 3, in which Λi = [ρi (ωi1 , t), ρi

(ωi2 , t), · · · , ρi (ωi J , t)]T and L i = diag{ωi1li1 , ωi2 li2 , Hence, ṡs −ι = −λD β̄ sgn(s). Thus, it can be shown

·· · , ωi J li J } < 0. Thus, one can conclude Ψ = that

3

i=1 Ψi < 0. On the other hand, since that k > 0 and (1 − ι)λt 1−β̄

α, β, γ are nonnegative constants, one has −(αx 2 + s 1−ι (t) − s 1−ι (0) = − . (25)

(1 − β̄)Γ (1 − β̄)

(β + k)y 2 + γ z 2 ) < 0. Therefore, one can conclude

Since s(treach ) = 0 at t = treach (where treach

dV (t)

= Ψ − (αx 2 + (β + k)y 2 + γ z 2 ) (22) denotes reaching time), one has

dt

1/1−β̄

(1 − β̄)Γ (1 − β̄)s 1−ι (0)

is negative. Thus, the FO nonlinear sliding mode treach = . (26)

dynamics (9) is asymptotically stable. The proof is (1 − ι)λ

completed.

(2) When the initial condition s(t) is less than zero

Next, the reachability condition of the sliding sur- (s(0) < 0), one has

face will be analyzed. ṡ(t) = −λ(−s)ι D β̄ sgn(s) ⇒ ṡ(−s)−ι

Theorem 4.2 Consider the FO nonlinear chaotic sys- = −λD β̄ sgn(s). (27)

tem (6) and switching surface (7), the state trajectories Thus, the integration of this equation can be

via the FO SMC u(t) (11) with the FO switching-type obtained as

control law (10) can converge to the s(t) = 0.

(1 − ι)λt 1−β̄

Proof Select a Lyapunov Candidate VS MC = s 2 (t). (−s(t))1−ι − (−s(0))1−ι = − .

(1 − β̄)Γ (1 − β̄)

Hence, the time derivative of VS MC is given by

(28)

V̇S MC = 2s(t)[D q2 y(t) + ϕ(t)]

= 2s(t)[g(x, y, z) + u(t) + x · f (x, y, z) Therefore, since s(treach ) = 0 at t = treach , the reaching

time can be shown as

+ z · h(x, y, z) + ky]

= 2s(t)[g(x, y, z) − g(x, y, z) − x · f (x, y, z) 1/1−β̄

(1 − β̄)Γ (1 − β̄)(−s(0))1−ι

− z · h(x, y, z) treach = . (29)

(1 − ι)λ

123

C. Yin et al.

1

|s(0)|1−ι (1 − β̄)Γ (1 − β̄)|s(0)|1−ι 1−β̄

Reaching time treach

(1 − ι)λ (1 − ι)λ

that

1/1−β̄

(1 − β̄)Γ (1 − β̄)|s(0)|1−ι

treach = . (30)

(1 − ι)λ

law u sw = −λ|s|ι sgn(s) (i.e., β̄ = 0), one has

s ṡ = −λ|s|ι ssgn(s) ≤ 0. There are two cases: (3) when

s(0) > 0, ṡ(t) = −λs ι −ι

. Hence, one has that ṡs = −λ.

Furthermore, s 1−ι = −(1 − ι)λ. Thus, one obtains

s 1−ι (t) = s 1−ι (0) − (1 − ι)λt. Since s(treach ) = 0, one

has treach = s 1−ι (0)/(1 − ι)λ. (4) When s(0) < 0,

−ṡ(−s)−ι = −λ. One has (−s)

1−ι

1−ι = −λ. So Fig. 1 Changes in treach along with varying β̄ and λ when

|s(0)| = 1.2, ι = 0.3, for the FO switching-type control law

one has (−s) 1−ι = −(1 − ι)λ. The integration of

this equation is given as (−s(t))1−ι = (−s(0))1−ι − ⎧ q

⎨ D 1 x = y · f (x, y, z) + z · ϑ(x, y, z) − αx,

(1 − ι)λt. Since s(treach ) = 0, one has treach =

D q2 y = g(x, y, z)−βy +Δg(x, y, z)+w(t)+u(t),

(−s(0))1−ι /(1 − ι)λ. According to the above cases (3) ⎩ q3

D z = y · h(x, y, z) − x · ϑ(x, y, z) − γ z,

and (4), one has

(32)

treach = |s(0)|1−ι /(1 − ι)λ. (31)

where Δg(x, y, z) and w(t) denote an uncertainty and

Remark 4.3 The calculation formulae of treach under an external disturbance. First, the following assump-

FO and IO switching-type control laws are listed in tion about the uncertainty and external disturbance is

Table 2. Due to D 0 sgn(s) = sgn(s), it is more likely for considered.

the FO switching-type control law to obtain better per-

Assumption 4.1 It is assumed that there exist two

formance. According to the calculation formula of treach

known d1 > 0 and d2 > 0 such that |Δg(x, y, z)| <

in (30), the parameters s(0), ι, λ and β̄ have influence

d1 , |w(t)| < d2 .

on treach . Figure 1 shows the changes in treach along with

λ and β̄, in which λ ∈ [2, 6] and β̄ ∈ [0, 0.98], when In order to make the state trajectories reach the slid-

|s(0)| = 1.2 and ι = 0.3. It shows that treach under the ing surface, a redefined FO switching-type control law

FO switching-type control law with 0 < β̄ ≤ 0.98 is is proposed as follows

smaller than the IO law does, when 2 ≤ λ ≤ 6. For the

other conditions, the smaller treach can be derived by

u sw = −(sgn(s)) − λ|s|ι D β̄ (sgn(s)), (33)

analyzing relationship between s(0), λ, ι, β̄ and treach ,

similar to the above case.

where > 0, 0 ≤ β̄ < 1, λ > 0 and 1 > ι > 0. From

Next, let us consider the 3D FO nonlinear system (6) (7) and (33), the overall SMC is presented by

in the presence of an uncertainty and an external dis-

u = −g(x, y, z) − x · f (x, y, z) − z · h(x, y, z) − ky

turbance. Let us modify description of the FO system

(9) as − sgn(s) − λ|s|ι D β̄ (sgn(s)). (34)

123

Adaptive fractional-order switching-type control method design

Theorem 4.3 Consider the uncertain 3D FO nonlin- Theorem 4.4 Consider the uncertain 3D FO nonlin-

ear system (32) and the switching surface (7), the ear system (32), this system is controlled by u(t) in

state trajectories via the FO SMC u(t) with the FO (37) with the adaptive laws in (36). Then, the corre-

switching-type control law can converge to s(t) = 0, if sponding state trajectories will converge to the sliding

> d1 + d2 . surface s(t) = 0.

Proof Set a Lyapunov candidate VS MC = s 2 (t). Proof Selecting a Lyapunov candidate V (t) = s 2 +

Hence, one has 1 1

(d̃1 − d̂1 )2 + (d̃2 − d̂2 )2 and taking its derivative,

V̇S MC = 2s(t)[D q2 y(t) + ϕ(t)] l1 l2

one has

= 2s(t)[D q2 y(t) + x · f (x, y, z)

V̇ = 2 s ṡ + (d̃1 − d̂1 )d̃˙ 1 + (d̃2 − d̂2 )d̃˙ 2

1 1

+ z · h(x, y, z) + βy + ky] l1 l2

= 2s(t)[g(x, y, z) + Δg(x, y, z) + w(t)

= 2 s(D q2 y(t) + ϕ(t)) + (d̃1 − d̂1 )d̃˙ 1

1

+ u(t) + x · f (x, y, z) l1

+ z · h(x, y, z) + ky]

+ (d̃2 − d̂2 )d̃˙ 2

1

≤ 2s(t)[|Δg| + |w(t)| − sgn(s) l2

= 2 {s(g(x, y, z) + Δg(x, y, z) + w(t) + u(t)

−λ|s|ι D β̄ (sgn(s))]. (35)

+ x · f (x, y, z) + z · h(x, y, z) + ky)

Since |Δg(x, y, z)| < d1 , |w(t)| < d2 , one can con-

+ (d̃1 − d̂1 )d̃˙ 1 + (d̃2 − d̂2 )d̃˙ 2 .

1 1

clude that VS MC > 0, V̇S MC < 0 from > d1 + d2 . (38)

Therefore, the inequality (35) confirms that the system l1 l2

(32) under the sliding mode controller (34) is globally Adding the adaptive laws (36) into the right side of

asymptotically stable.

(38), one can conclude

Remark 4.4 In order to use the controller (34), the V̇ = 2{s(g(x, y, z) + Δg(x, y, z) + w(t) + u(t)

bound d1 and d2 of Δg(x, y, z) and w(t) should be + x · f (x, y, z) + z · h(x, y, z) + ky)

known. However, the functions in real world are fully + (d̃1 − d̂1 )|s| + (d̃2 − d̂2 )|s|}. (39)

unknown. To overcome this, the following assumption

and the control laws are proposed. From the controller (37), this yields

Assumption 4.2 The unknown uncertainty Δg(x, y, z) V̇ = 2{s[Δg(x, y, z) + w(t)] − s[(d̃1 + d̃2 )sgn(s)

and the external disturbance w(t) are assumed to + λ|s|ι D β̄ sgn(s)]

be bounded. Therefore, there exist appropriate posi- + (d̃1 − d̂1 )|s| + (d̃2 − d̂2 )|s|}. (40)

tive constants d̂1 and d̂2 such that |Δg(x, y, z), | <

d̂1 , |w(t)| < d̂2 . Furthermore, the constants d̂1 and It is apparent that

d̂2 are unknown. V̇ ≤ 2{|s|[|Δg(x, y, z)|+|w(t)|] − s[(d̃1 + d̃2 )sgn(s)

To handle Δg(x, y, z) and w(t), appropriate adap- + λ|s|ι D β̄ (sgn(s))] + (d̃1 − d̂1 )|s|

tive laws can be defined as follows + (d̃2 − d̂2 )|s|}. (41)

By using Assumption 4.2, one can derive

d̃˙ 1 = l1 |s|, d̃1 (0) = d̃10 , d̃˙ 2 = l2 |s|, d̃2 (0) = d̃20 ,

V̇ ≤ 2{|s|(d̂1 + d̂2 ) − s[(d̃1 + d̃2 )sgn(s)

(36)

+ λ|s|ι D β̄ (sgn(s))]

in which d̃1 and d̃2 are estimations for d̂1 and d̂2 , respec- + (d̃1 − d̂1 )|s| + (d̃2 − d̂2 )|s|}

tively. l1 and l2 are positive constants and d̃10 and d̃20 = 2{−s[(d̃1 + d̃2 )sgn(s) + λ|s|ι D β̄ sgn(s)]

are the initial conditions. Therefore, the overall control

+ d̃1 |s| + d̃2 |s|}.

law is modified from Assumption 4.2 and (36)

u(t) = −g(x, y, z) − x · f (x, y, z) − z · h(x, y, z) Changing ssgn(s) into |s|, it is clear that

−ky − (d̃1 + d̃2 )sgn(s) − λ|s|ι D β̄ (sgn(s)). V̇ ≤ 2{−(d̃1 + d̃2 )|s|−λ|s|ι s D β̄ sgn(s)+ d̃1 |s|+ d̃2 |s|}

(37) = −2λ|s|ι s D β̄ sgn(s). (42)

123

C. Yin et al.

Thus, from Lemma 2.1, similar to the above analysis,

the system trajectories can arrive at the predefined slid-

ing surface.

70

60

50

5 Numerical examples 40

z(t)

30

20

In this section, our goal is to achieve SMC by apply- 10

40

systems. 30

20

10

Example 5.1 Consider FO Lorenz system given by 0

20

30

⎧

−10

y(t) 10

f α −20 0

⎪

⎪ −10

⎪ x(t)

−30

⎪ D x = a y − a x,

q −20

⎪

1 −40 −30

⎪

⎪

⎪

⎪

g

β

⎪

⎨ q2 Fig. 2 Chaotic trajectories of FO Lorenz system

D y = x(b − z) − 1 y + u(t)

Δg w(t)

(43)

⎪

⎪

⎪

⎪

⎪

⎪ + 0.07 sin(0.15π y) + 0.07 sin(0.15t), u I O = −bx −ax − y −0.14sgn(s)−3.5sgn(s), (46)

⎪

⎪

⎪

⎪ h γ

u I O1 = −bx − ax − y − 0.14sgn(s) − sgn(s), (47)

⎩ q

D z = x y − c z,

3

where (a, b, c) = (10, 38, 28/3) and q1 = 0.985, q2 = type control law. The comparisons between u in (45)

0.993, q3 = 0.99. In the absence of the uncertainty and with the FO switching-type control law and the corre-

external disturbance, the system (43) without the con- sponding SMC u I O and u I O1 with the IO reaching law

troller exhibits a chaotic behavior as shown in Fig. 2. are depicted in Figs. 4, 5 and 6. Part (a) of Fig. 4 shows

First, from (8) and (34), the switching surface and the comparisons between x under u, u I O and u I O1 .

the SMC with the FO switching-type control law are Part (b) of Fig. 4 shows the comparisons between y

derived as under u, u I O and u I O1 . Part (c) of Fig. 4 shows the

t

q2 −1 comparisons between z under u, u I O and u I O1 . The

s=D y+ {ax(τ ) + x(τ )z(τ ) + 2y(τ )}dτ,

0 corresponding sliding surfaces under u, u I O and u I O1

(44) are presented in Fig. 5. The time responses of u, u I O

u = −bx − ax − y − 0.14sgn(s) and u I O1 are drawn in Fig. 6. They show that the results

under the proposed FO switching-type control law own

−3.5|s|0.78 D 0.25 sgn(s). (45)

a faster and lower chattering performance.

The initial condition is [x(0), y(0), z(0)]T = [−1.37,

3, −2]T . The fractional integration operator is approxi-

Example 5.2 Consider FO Rucklidge oscillator

mated via Carlson method in frequency range (0.01,100)

rad/s by using MATLAB toolbox called Ninteger. The ⎧

⎪

⎪

f

α

states x, y and z under the controller (45) are drawn ⎪

⎪

⎪

⎪ D x = (b − z) y − a x,

q 1

in Part (a) of Fig. 3. The sliding surface (44) and the ⎪

⎪

⎪

⎪ g Δg

SMC (45) are separately depicted in Part (b) and Part ⎪

⎨ q2

(c) of Fig. 3. They show that the SMC (45) with the D y = x +u(t) + 0.1 sin(0.2π y)

(48)

FO switching-type control law guarantee the reaching ⎪

⎪

w(t)

⎪

⎪

to the sliding surface and finally stabilizing. ⎪

⎪ + 0.1 sin(0.2t),

⎪

⎪

⎪

⎪ γ

Next, in order to show the better control perfor- ⎪

⎩ q3

h

mance of the proposed control law, some comparisons D z = y y − 1 z,

are shown. From the IO reaching law in [22], the cor-

responding FO SMC with the traditional IO reaching where (a, b) = (2, 10.7) and q1 = q2 = q3 = 0.95.

law is obtained as The system (48) exhibits a chaotic behavior with-

123

Adaptive fractional-order switching-type control method design

of stabilized FO Lorenz

system under the controller x

2 y

(45); b the time response of

the sliding surface (44); z

1

c the time response of the

x,y,z

FO SMC (45)

0

−1

−2

0 1 2 3 4 5 6 7

Time(sec)

(b) (c)

4 40

s u

3 20

2 0

s

u

1 −20

0 −40

−1 −60

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

Time(sec) Time(sec)

the states x, y and z under 1.5

the proposed controller (45) x

(blue line); time responses 1 x

IO

of the states x, y and z under x

0.5 IO1

u I O (red line) and time

responses of the states x, y 0

and z under u I O1 (green

line). (Color figure online) −0.5

−1

−1.5

0 1 2 3 4 5 6 7

Time(sec)

(b) (c)

3 0.5

y

2 y 0

IO

y z

IO1

1 −0.5

z

IO

0 −1 zIO1

−1 −1.5

−2 −2

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

Time(sec) Time(sec)

123

C. Yin et al.

5 s

sIO

sIO1

4 25

Sliding surface

20

3

15

z(t)

2 10

1

0

−5

0 10

5 20

0 1 2 3 4 5 6 7 15

0 10

Time(sec) 5

0

−5 −5

y(t) −10 x(t)

Fig. 5 Time response of the sliding surface s under the proposed −10 −15

s I O under u I O (red line) and time response of the sliding surface Fig. 7 Chaotic trajectories of FO Rucklidge oscillator

s I O1 under u I O1 (green line). (Color figure online)

60

u (51)

uIO

40 uIO1

Control input

20

as the initial condition. The states x, y and z under

the controller (50) are drawn in Part (a) of Fig. 8. The

0

sliding surface (49) and the FO SMC (50) are separately

−20 depicted in Part (b) and Part (c) of Fig. 8. They show

that the FO SMC (49) guarantees the reaching to the

−40 sliding surface and finally stabilizing.

Next, in order to show the better control perfor-

−60 mance of the proposed control law, some comparisons

0 1 2 3 4 5 6 7

Time(sec) are shown. From the adaptive law in [24], the cor-

responding SMC with the IO reaching law is given

Fig. 6 Time response of the proposed controller (45) (blue line); as

time response of u I O (red line) and time response of u I O1 (green

line). (Color figure online) u I O = −x − x(b − z) − zy − y

out Δg, w and the controller, which is depicted in − (1 + d̃1 + d̃2 )sgn(s), (52)

Fig. 7.

The comparisons between the SMC (50) with the FO

First, from (8) and (37), the switching surface and the switching-type control law and the corresponding SMC

SMC with FO switching-type control law are written u I O described in (52) are depicted in Figs. 9, 10

as and 11. Part (a) of Fig. 9 shows the comparisons

t between x under u and x I O under u I O . Part (b)

s(t) = D q2 −1 y + {x(b − z) + zy + y}dτ, (49) of Fig. 9 shows the comparisons between y under

0 u and u I O . Part (c) of Fig. 9 shows the compar-

u(t) = −x − x(b − z) − zy − y − |s|0.67 D 0.21 sgn(s) isons between z under u and u I O . The correspond-

− (d̃1 + d̃2 )sgn(s), (50) ing sliding surfaces under u and u I O are presented in

Fig. 10. The time responses of u and u I O are drawn in

with the adaptive laws Fig. 11. Figures 9, 10 and 11 show that the obtained

123

Adaptive fractional-order switching-type control method design

of stabilized FO Rucklidge

2

system under the controller

(50); b the time response of 1

the sliding surface (49); c 0

the time response of the FO

x,y,z

SMC (50) −1 x

−2 y

z

−3

−4

0 1 2 3 4 5 6 7

Time(sec)

(b) (c)

1 50

s u

0 40

30

−1

20

−2

s

u

10

−3

0

−4 −10

−5 −20

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

Time(sec) Time(sec)

the states x, y and z under 2

the proposed controller (50) x

(red line) and time 1 x

IO

responses of the states x, y

0

and z under u I O (green −4

x 10

line). (Color figure online) −1

5

−2

−3 0

−4

−5

5.2 5.6 6 6.4

−5

0 1 2 3 4 5 6 7

Time(sec)

(b) (c)

3 0

y z

2 y

IO zIO

−1

1

0 −2

−3

x 10

2

−1

0 −3

−2

−2

5.2 5.6 6 6.4

−3 −4

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

Time(sec) Time(sec)

123

C. Yin et al.

1

S Appendix: Proof of Lemma 2.1

SIO

0

For ∀t > 0, there exists a time interval (tk , tk+1 ] such

−3

x 10 that t ∈ (tk , tk+1 ] and σ (t ) ≥ 0, ∀t ∈ (tk , tk+1 ] if

Sliding surface

−1

1

σ (t) > 0, or σ (t ) ≤ 0, ∀t ∈ (tk , tk+1 ] if σ (t) < 0.

0.5

−2 Furthermore, there exists a finite partition given by

0

0 = t0 < t1 < t2 < · · · < tk−1 < tk , such that:

−3 −0.5

1) for every interval (ti , ti+1 ], (i = 0, 1, · · · , k − 1),

σ (t ) ≤ 0, ∀t ∈ (ti , ti+1 ] or σ (t ) ≥ 0, ∀t ∈ (ti , ti+1 ];

−1

4.4 4.6 4.8 5 5.2

−4

and 2) σ (t ) ≤ 0, ∀t ∈ (ti+1 , ti+2 ] if σ (t ) ≥ 0, ∀t ∈

−5 (ti , ti+1 ] or σ (t ) ≥ 0, ∀t ∈ (ti+1 , ti+2 ] if σ (t ) ≤

0 1 2 3 4 5 6 7

0, ∀t ∈ (ti , ti+1 ], for every two adjacent intervals

Time(sec)

(ti , ti+1 ] and (ti+1 , ti+2 ]. Moreover, we require that the

Fig. 10 Time response of the sliding surface s under the pro- initial time in every (ti , ti+1 ] is not equal to zero. From

posed controller (50) (red line) and time response of the sliding the integral properties, denoting t0 = 0, one has

surface s I O under u I O (green line). (Color figure online)

β̄ [ f 0 (t) + f 1 (t) + · · · + f k (t)]

60 Dt sgn(σ (t)) = , (53)

u

uIO

Γ (1 − β̄)

50 3

2

d ti+1 sgn(σ (τ ))

40

1 where f i (t) = dτ , (i = 0, 1, 2, · · · ,

dt ti (t−τ )β̄

Control input

30 0

d t sgn(σ (τ ))

k − 1), f k (t) = dt tk dτ .

(t−τ )β̄

−1

20 −2

First, we consider σ (t) > 0. From the above analy-

sis, one has σ (t ) ≥ 0, ∀t ∈ (tk , tk+1 ]. There exists

−3

4.4 4.6 4.8 5 5.2

10

0 tk = tk0 < tk1 < tk2 < · · · < tklk −1 < tklk = t in

(tk , t] such that (tk , t] = (tk0 , tk1 ] ∪ (tk1 , tk2 ] ∪ · · · ∪

(tklk −1 , tklk ]. Moreover, σ (t ) ≥ 0, ∀t ∈ (tk0 , tk1 ] in

−10

−20

0 1 2 3 4 5 6 7

which certain σ (t ) = 0 just happen at some isolate

Time(sec) points t ; σ (t ) ≡ 0, ∀t ∈ (tk1 , tk2 ]; σ (t ) ≥ 0, ∀t ∈

(tk2 , tk3 ] in which certain σ (t ) = 0 just happen at some

Fig. 11 Time response of the proposed controller (50) (red line) isolate points t ; · · · ; σ (t ) ≡ 0, ∀t ∈ (tklk −2 , tklk −1 ];

and time response of u I O (green line). (Color figure online) σ (t ) ≥ 0, ∀t ∈ (tklk −1 , tklk ] in which σ (t ) = 0 just

happen at isolate points t . In addition, we also require

theoretic results can have the better control perfor- that the initial time in (tk j , tk j+1 ], ( j = 0, 1, · · · , lk −

mance. 1) is not zero. Thus, one has

d t sgn(σ (τ ))

dτ = (t − tklk −1 )−β̄ ,

6 Conclusion dt tklk −1 (t − τ )β̄

tkl −1

d k sgn(σ (τ ))

In this paper, the FO switching-type control law has dτ

dt tklk −2 (t − τ )β̄

been introduced and the adaptive SMC has been

designed to guarantee the robust stability in the FO non- d tklk −1 0

= dτ = 0,

linear system. The calculation formula of the reaching dt tklk −2 (t − τ )β̄

time has been analyzed and computed. Stability analy- tklk −2 sgn(σ (τ ))

d

sis of the FO sliding mode dynamics has been derived. dτ = (t − tklk −3 )−β̄

dt tklk −3 (t − τ )β̄

The proposed theoretic results based on the FO sign

function can have a faster and lower chattering perfor- −(t − tklk −2 )−β̄ ,

mance. Finally, numerical examples have shown the

advantage of the proposed control scheme. and so on, one can conclude

123

Adaptive fractional-order switching-type control method design

f k (t) = (t − tk0 )−β̄ − (t − tk1 )−β̄ + (t − tk2 )−β̄ [ f 0 (t) + f 1 (t) + · · · + f k (t)]

D β̄ sgn(σ (t)) = > 0.

Γ (1 − β̄)

−(t − tk3 )−β̄ + · · · + (t − tklk −1 )−β̄ . (54)

(57)

Since (t − tk j )−β̄ is an increasing function in tk j ,

f k (t) ≥ (t − tk )−β̄ . Then, we discuss f i (t). When Next, we consider the second case when σ (t) < 0.

σ (t ) ≥ 0, ∀t ∈ (ti , ti+1 ], there exists a time partition Similar to the first case, we have

(ti , ti+1 ] = (ti0 , ti1 ] ∪ (ti1 , ti2 ] ∪ · · · ∪ (tili −1 , tili ] in

which ti0 = ti , tili = ti+1 . Moreover, σ (t ) ≥ 0, ∀t ∈ [ f 0 (t) + f 1 (t) + · · · + f k (t)]

D β̄ sgn(σ (t)) = < 0.

(ti0 , ti1 ] in which σ (t ) = 0 just happen at some isolate Γ (1 − β̄)

points t ; σ (t ) ≡ 0, ∀t ∈ (ti1 , ti2 ]; σ (t ) ≥ 0, ∀t ∈ (58)

(ti2 , ti3 ] in which certain σ (t ) = 0 just happen at

some isolate points t ; and so on. We also claim that

the initial time in (ti j , ti j+1 ], ( j = 0, 1, · · · , li − 1) is References

not zero. Considering (tili −1 , tili ], there are two possi-

bilities (i.e., σ (t ) ≥ 0, ∀t ∈ (tili −1 , tili ] or σ (t ) ≡ 1. Podlubny, I.: Fractional Differential Equations. Academic

0, ∀t ∈ (tili −1 , tili ]). Hence, f i (t) can be calculated, Press, New York (1999)

similarly the calculation of f k (t), f i (t) ≥ (t − ti )−β̄ − 2. Ahn, H.S., Chen, Y.Q.: Necessary and sufficient stability

condition of fractional-order interval linear systems. Auto-

(t − ti+1 )−β̄ , or f i (t) ≥ (t − ti0 )−β̄ − (t − tili −1 )−β̄ . matica 44, 2985–2988 (2008)

When σ (t ) ≤ 0, ∀t ∈ (ti , ti+1 ], there exists 3. Li, Y., Chen, Y.Q., Podlubny, I.: Mittag–Leffler stability of

(ti , ti+1 ] such that (ti , ti+1 ] = (ti0 , ti1 ] ∪ (ti1 , ti2 ] ∪ fractional order nonlinear dynamic systems. Automatica 45,

1965–1969 (2009)

· · · ∪ (tili −1 , tili ] in which ti0 = ti , tili = ti+1 . More-

4. Maione, G.: Conditions for a class of rational approximants

over, σ (t ) ≤ 0, ∀t ∈ (ti0 , ti1 ] in which σ (t ) = 0 of fractional differentiators/integrators to enjoy the interlac-

just happen at some isolate points t ; σ (t ) ≡ 0, ∀t ∈ ing property. In: Proceeding of the 18th IFAC World Congr,

(ti1 , ti2 ]; σ (t ) ≤ 0, ∀t ∈ (ti2 , ti3 ] in which cer- 18, 13984–13989 (2011)

tain σ (t ) = 0 just happen at some isolate points 5. Trigeassou, J.C., Maamri, N., Sabatier, J., Oustaloup, A.:

State variable and transients of fractional order differential

t ; and so on. We also claim that the initial time in systems. Comput. Math. Appl. 64(10), 3117–3140 (2012)

(ti j , ti j+1 ], ( j = 0, 1, · · · , li −1) is not zero. Consider- 6. Li, C., Ma, Y.: Fractional dynamical system and its lineariza-

ing (tili −1 , tili ], there are two possibilities (i.e., σ (t ) ≤ tion theorem. Nonlinear Dyn. 71, 621–633 (2013)

0, ∀t ∈ (tili −1 , tili ] or σ (t ) ≡ 0, ∀t ∈ (tili −1 , tili ]). 7. Maione, G.: On the Laguerre rational approximation to frac-

tional discrete derivative and integral operators. IEEE Trans.

Hence, one can conclude that Automat. Control 58, 1579–1585 (2013)

8. Sun, H., Chen, W., Chen, Y.Q.: Variable-order fractional dif-

d ti1 sgn(σ (τ ))

dτ = (t − ti1 )−β̄ − (t − ti0 )−β̄ , ferential operators in anomalous diffusion modeling. Phys.

dt ti0 (t − τ )β̄ A 388(21), 4586C92 (2009)

(55) 9. Machado, J.A.T.: Fractional order modelling of fractional-

order holds. Nonlinear Dyn. 70, 789–796 (2012)

d ti2 sgn(σ (τ )) d ti2 0

dτ = dτ = 0, 10. Monje, C.A., Chen, Y.Q., Vinagre, B.M., Xue, D., Feliu, V.:

dt ti1 (t − τ )β̄ dt ti1 (t − τ )β̄ Fractional order Systems and Controls: Fundamentals and

Applications. Springer, London, New York (2010)

(56)

11. Luo, Y., Chen, Y.Q., Pi, Y.: Experimental study of fractional

and so on, one has order proportional derivative controller synthesis for frac-

tional order systems. Mechatronics 21, 204–214 (2011)

f i (t) = (t − ti1 )−β̄ −(t − ti0 )−β̄ +· · · + (t − tili )−β̄ 12. Yin, C., Stark, B., Chen, Y.Q., Zhong, S.M.: Adaptive min-

imum energy cognitive lighting control: integer order vs

−(t − tili −1 )−β̄ > 0, fractional order strategies in sliding mode based extremum

seeking. Mechatronics 23, 863–872 (2013)

or 13. Li, R., Chen, W.: Lyapunov-based fractional-order con-

troller design to synchronize a class of fractional-order

f i (t) = (t − ti1 )−β̄ −(t − ti0 )−β̄ +· · ·+(t −tili −1 )−β̄ chaotic systems. Nonlinear Dyn. 76, 785–795 (2014)

14. Yin, C., Chen, Y.Q., Zhong, S.M.: Fractional-order sliding

−(t − tili −2 )−β̄ > 0. mode based extremum seeking control of a class of nonlinear

system. Automatica 50, 3173–3181 (2014)

k

15. Utkin, V.I.: Sliding Modes in Control and Optimization.

So, one can conclude f i (t) > 0. Thus, we have

i=0 Springer, New York (1992)

123

C. Yin et al.

16. Liu, J.K., Wang, X.H.: Advanced Sliding Mode Control for 24. Yin, C., Dadras, S., Zhong, S., Chen, Y.Q.: Control of a

Mechanical Systems: Design, Analysis and MATLAB Sim- novel class of fractional-order chaotic systems via adaptive

ulation. Springer, Tsinghua University Press, Berlin, Beijing sliding mode control approach. Appl. Math. Modell. 37(4),

(2012) 2469–2483 (2013)

17. Tavazoei, M.S., Haeri, M.: Chaotic attractors in incommen- 25. Tian, X., Fei, S.: Robust control of a class of uncertain

surate fractional order systems. Phys. D 237, 2628–2637 fractional-order chaotic systems with input nonlinearity via

(2008) an adaptive sliding mode technique. Entropy 16, 729–746

18. Hosseinnia, S.H., Ghaderi, R., Ranjbar, N.A., Mahmoudian, (2014)

M., Momani, S.: Sliding mode synchronization of an uncer- 26. Aghababa, M.P.: A novel terminal sliding mode controller

tain fractional order chaotic system. Comput. Appl. Math. for a class of non-autonomous fractional-order systems.

59, 1637–1643 (2010) Nonlinear Dyn. 73, 679–688 (2013)

19. Odibat, Z.M.: Adaptive feedback control and synchroniza- 27. Efe, M.: Fractional fuzzy adaptive sliding-mode control of

tion of non-identical chaotic fractional order systems. Non- a 2-DOF direct-drive robot arm. IEEE Trans. Syst. Man

linear Dyn. 60, 479–487 (2010) Cybern. Part B Cybern. 38, 1561–1570 (2008)

20. Tavazoei, M.S., Haeri, M.: Synchronization of chaotic 28. Yin, C., Stark, B., Chen, Y.Q., Zhong, S.M., Lau, E.:

fractional-order systems via active sliding mode controller. Fractional-order adaptive minimum energy cognitive light-

Phys. A 387, 57–70 (2008) ing control strategy for the hybrid lighting system. Energy

21. Wang, X., Zhang, X., Ma, C.: Modified projective synchro- Build. 87, 176–184 (2015)

nization of fractional-order chaotic systems via active slid- 29. Yin, C., Chen, Y.Q., Zhong, S.M.: Fractional-order power

ing mode control. Nonlinear Dyn. 69, 511–517 (2012) rate type reaching law for sliding mode control of uncertain

22. Yin, C., Zhong, S., Chen, W.: Design of sliding mode con- nonlinear system. In: Proceeding of 19th International Fed-

troller for a class of fractional-order chaotic systems. Com- eration of Automatic Control World Congress, Cape Town,

mun. Nonlinear Sci. Numer. Simul. 17, 356–366 (2012) South Africa, 5369–5374 (2014)

23. Zhang, R., Yang, S.: Robust synchronization of two different 30. Trigeassou, J.C., Maamri, N., Sabatier, J., Oustaloup, A.: A

fractional-order chaotic systems with unknown parameters Lyapunov approach to the stability of fractional differential

using adaptive sliding mode approach. Nonlinear Dyn. 71, equations. Signal Process. 91, 437–445 (2011)

269–278 (2013)

123

- CL7103_SystemTheoryquestionbankUploaded bysyed1188
- Li&Kumar2011revisedUploaded byAfshin
- CL7103-System Theory QBUploaded bysyed1188
- Control of Quadrotor Using Sliding Mode Disturbance Observer and Nonlinear H-InfiniteUploaded byInternational Journal of Robotics, Theory and Applications
- A Mechatronic Analysis and Synthesis of Human Walking GaitUploaded byEdgar Guerrero
- Adaptive Integral-type Sliding Mode Control for Spacecraft AttitudeUploaded byoscar201140
- ABS Modelling Ref Paper (1)Uploaded byashok b v
- 1-s2.0-S0005109806004432-main.pdfUploaded bysethukumark
- 125_600Uploaded byBhaskar Biswas
- 06Nov2012121035Sliding Mode Controllers....Uploaded byLina Al-Saleh
- detectability subspacesUploaded byGhaltagh MMM
- Springer Atif.sanjeev 2017Uploaded byMurali Rv
- Mozell i 2009Uploaded bySaikat Gope
- Stochastic Stability PropertiesUploaded byJosiane Ferreira
- Adaptive controlUploaded byHamza Masre
- Industrial View on Control Teaching TheoryUploaded byAshrul Aizaed
- Advanced PID Control_AstromHagglund_Chapter1 IntroductionUploaded byJordi El Mariachet
- el07-106Uploaded bySenthil Kumar
- PDF v f Using PID and Encoder EnUploaded byacount.br
- SCE-Advanced PID Control S7-1200 (2016)Uploaded byJorge Andril
- Decentralized Design of Feedback Control of Large Scale SystemUploaded byBinhMinh Nguyen
- siUploaded bySumi Srianand
- 03-System Identification and Control of an Electro-Hydraulic Actuator SystemUploaded byDaniel Infante
- 1Uploaded byMohamad Abdul Raya
- PID_Control & Ziegler-NicholsUploaded byHachemi Bhe
- A Tutorial on Preview Control SystemsUploaded byjaquetona
- 03 Cme LyapunovUploaded bySulung Fitrianto
- heating_failure_detection.pdfUploaded byHugh Manta
- HA026933_9_2704EngUploaded bynvdung0311
- IJAIEM-2014-05-22-063Uploaded byAnonymous vQrJlEN

- sol2Uploaded bysalim
- tenta13A_2Uploaded bysalim
- NLSC Lecture Notes 2016 Sufficient Conditions ChapterUploaded bysalim
- IOIPNotesOct2014Uploaded bysalim
- Ma 30087Uploaded bysalim
- 08520064Uploaded bysalim
- 511 NotesUploaded bysalim
- MA30087/50087: Optimisation methods of operational researchUploaded byMarcelo Ziulkoski
- TP2_4Uploaded bysalim
- t 2Uploaded bysalim
- t3Uploaded bysalim
- t 1Uploaded bysalim
- Tentalosning TMA947 070312 2Uploaded bysalim
- intro_optUploaded byУзахир УЗИ Армани
- adap 1Uploaded bysalim
- EEE482-FA09-ISSMUploaded bysalim
- 07 Homework6 SolUploaded bysalim
- tentoct11Uploaded bysalim
- lecture_notes_series_optimisation_constrained.pdfUploaded bysalim
- cv-14-lecture-18_3Uploaded bysalim
- hu2008Uploaded bysalim
- Lewis 1996Uploaded bysalim
- m11_lec1Uploaded bysalim
- SEEM3470_tuto_02Uploaded bysalim
- intro_optUploaded byУзахир УЗИ Армани
- intro_optUploaded byУзахир УЗИ Армани
- intro_optUploaded byУзахир УЗИ Армани
- intro_optUploaded byУзахир УЗИ Армани
- Course Outlines 012Uploaded bysalim

- Nucleo-F207ZGUploaded byMani Kandan K
- RoboticsUploaded byapi-3707156
- cm6800 32x16Uploaded byvijayakumar3661
- An Efficient Approach of Fast External Sorting Algorithm in Data WarehouseUploaded byJournal of Computing
- Gate MathsUploaded byAyush Maheshwari
- Simulate ONTAP 9.3 Installation and Setup GuideUploaded bysushant_beura
- Online Examination SystemUploaded byNisha Nair
- C++ ProgrammingUploaded bymarius_v3n0m5417
- Oracle NotesUploaded byvaaasu
- ymrtc_log.txtUploaded byVinicius Silveira
- Converting ZeroShell 2GB Disk Image to a 1GB Disk Image With InstallationUploaded byAgus Kresna Ardiyana
- napkin problemUploaded byDian Septiana
- Tender Document RcsUploaded byAnshika Jain
- Operations Manual Cs3 RegiusUploaded byRoyMezaRivaldo
- ERP Gantt Chart-MMUploaded byashfaqmemon2001
- CMPE 287&187 Homework 1 Spring 2012Uploaded bywhoisnull
- Smart CityUploaded byAjirAja
- DataBank Manual BookUploaded byalan_yaobin8409
- ABAP XMLUploaded bydpmallick
- Using boxes and glue in TEX and LATEXUploaded bymalekan2005
- Curso Control mUploaded byMartindelaFuente
- Critical Path in Network AnalysisUploaded byminoshini
- WP Qlik Sense Architectural OverviewUploaded byramiro.padin6634
- Juniper Networks Netscreen 204 110004Uploaded byapi-3775248
- 13 Electric Motor HandbookUploaded byWâhýûðî Fãjâr
- 5 Developing a Project Plan-IIUploaded byashisb444
- Introduction to Coded UI Tests With Visual Studio Ultimate 2012.Uploaded byMahesh Reddy Gade
- 11.0 Lampiran -Alternative Subtraction AlgorithmsUploaded by江沙阿炳
- Leighton Seager, 1st Baron Leighton of St MellonsUploaded byMoshe Rubin
- D1_Inventory_of_Methods_Risk_Management_Final.pdfUploaded byJuan Carlos Campos