This article was downloaded by: [Canadian Research Knowledge Network] On: 3 August 2010 Access details: Access

Details: [subscription number 918588849] Publisher Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK

International Journal of Control

Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713393989

On adaptive robust backstepping control schemes suitable for PM synchronous motors

Jian-Xin Xu Tong Heng Lee Qing-Wei Jia Mao Wang

To cite this Article Wang, Jian-Xin Xu Tong Heng Lee Qing-Wei Jia Mao(1998) 'On adaptive robust backstepping control

schemes suitable for PM synchronous motors', International Journal of Control, 70: 6, 893 920 To link to this Article: DOI: 10.1080/002071798222028 URL: http://dx.doi.org/10.1080/002071798222028

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

INT. J. CONTROL,

1998, VOL. 70, NO. 6, 893 920

On adaptive robust backstepping control schemes suitable for PM synchronous motors

JIAN-XIN XU , TONG HENG LEE , QING-WEI JIA and MAO WANG
This paper presents new adaptive robust control strategies suitable for permanent magnet (PM) synchronous motors. By using the backstepping approach, both the adaptive and robust controllers are appropriately designed to ensure global stability. Taking into account the variation ranges of system parameters, the adaptive robust methods are further developed to achieve better tracking performance. It is shown that the synthesized adaptive robust control schemes developed here can retain advantages of both adaptive and robust control schemes and overcome their shortcomings. To avoid discontinuous control laws which may cause problems in both theoretical and practical aspects, a continuous adaptive robust control method associated with a -modi cation scheme is also proposed to guarantee both the uniform boundedness of the system and suitably designated tracking precision. The paper includes simulation studies demonstrating the performance of the proposed control schemes.

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

1.

Introduction

Various adaptive control algorithms for nonlinear systems with parametric uncertainties have been proposed and developed over the last decade (Taylor et al. 1989, Sastry and Isidori 1989, Narendra and Annaswamy 1989, Kokotovic 1991). On the other hand, the problem of designing stabilizing control for nonlinear systems containing interval uncertainties has also been the central subject of research on robust control strategies (Gutman 1979, Corless and Leitmann 1981, Barmish et al. 1983, Slotine and Li 1992, Utkin 1992). More recently, by using backstepping approaches, adaptive and robust control methods have been extended respectively to cascaded nonlinear systems (Kanellakopoulos et al. 1991, Qu 1993, Jiang and Pomet 1995). There are also some reports on the synthesis of adaptive and robust control schemes, see Ioannou and Sun (1996) for instance. However, little has been done on the design of robust adaptive backstepping control for cascaded nonlinear systems in the presence of both parametric and interval uncertainties. The main objective of this paper is to develop new adaptive robust control schemes incorporating backstepping design approaches for a class of commonly encountered servo-actuators with nonlinear dynamics the permanent magnet synchronous motors. As a typical nonlinear control system, the eld of motor control is such a broad research area that many nonlinear control techniques can be applied to address the di erent motor control problems. For instance, nonlinear control such as feedback linearization, adaptive, robust and sliding mode control schemes have been well
Department of Electrical Engineering, National University of Singapore, 10 Kent Ridge
0020-7179/98$12.00 1998 Taylor & Francis Ltd.

Received 7 April 1997.

Crescent, Singapore 119260.

894

J.-X. Xu et al.

developed for motor control (Liao et al. 1990, Dawson et al. 1994, Fujimoto and Kawamura 1995). Adaptive robust control schemes with speci ed tracking error bounds were presented in Liao et al. (1990) and Brogliato and Neto (1995). However, those methods are only suitable for one-step design of the systems. In other words, their algorithms cannot be extended to the backstepping design which requires that the input be di erentiable at each step and consists of at least two steps. Through the use of ingenious ideas like integrator backstepping, it is now possible to design controllers for nonlinear electromechanical systems such as electric motors actuating a robotic load (Dawson et al. 1994). Since the permanent magnet synchronous motors are widely applied to drive fast dynamic loads, we restrict our interest to some of the recent nonlinear control work in this area for position or velocity tracking control problems. The dynamics of the permanent magnet synchronous motor can be presented by a dynamic electrical subsystem and a dynamic mechanical subsystem, which are nonlinear di erential equations. Strictly speaking, most control methods for permanent magnet synchronous motors are only locally stable because the d-axis current is assumed to be zero and the design procedure is based on the reduced model. In Fujimoto and Kawamura (1995), a robust control scheme based on two-degree-offreedom control with sliding mode is proposed for PM DC motors, where the delay of electrical response is ignored. In Dawson et al. (1994), both the adaptive backstepping and robust backstepping approaches are applied to permanent magnet synchronous motors which are of the parametric-pure feedback form, i.e. the controller design is based on the reduced model. It is well known that backstepping design methods essentially require the parametric-pure feedback form. In this paper, we apply backstepping approaches to design adaptive and robust controllers for permanent magnet synchronous motors in which the nonlinear dynamical equations are not exactly in the parametric-pure feedback form. Instead of only zeroing d-axis current, the extra d-axis control input voltage is used to deal with the nonlinear coupling part of the dynamics as well. Hence, the proposed controller ensures the strictly global stability of the whole electromechanical systems. Taking into account the range of system parametric variations, the adaptive and robust approaches are further synthesized to form an adaptive robust controller for the permanent magnet synchronous motor. The robust technique is used to suppress the motor parametric uncertainties which usually are much smaller than the load changes. Since it is di cult to give upper bound of the load variations and an overlarge estimate of the upper bound will result in excessive control authority, adaptive technique is adopted to tackle the payload uncertainty. By using switching functions, the proposed adaptive robust controller ensures global stability of the whole electromechanical system. However, incorporating discontinuous functions in the controller may give rise to problems related to the existence and uniqueness of solutions as well as problems in practical implementation. To deal with these problems, the controller with continuous functions instead of switching functions is further developed and analysed, which guarantees the uniform boundedness of the system tracking error. Moreover, The -modi cation scheme (Xu et al. 1997) is used to cease parameter adaptation in accordance with the adaptive robust control law. In this method, the tracking precision can be made arbitrarily small by choosing suitable control design parameters. This paper is organized as follows. Section 2 describes the dynamic model of permanent magnet synchronous motor. Section 3 gives the adaptive backstepping

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Control schemes for PM synchronous motors

895

design procedure and stability analysis. Section 4 presents the robust backstepping design method and relevant analysis. Section 5 details the synthesis of adaptive and robust control schemes using switching functions. Section 6 presents the adaptive robust backstepping design by using continuous functions to approximate the switching functions adopted in 5 as well as the -modi cation scheme. Simulation-based studies on the proposed methods are given in 7.

2.

M odel of permanent magnet synchronous motor

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

A permanent magnet synchronous motor is described by the following subsystems: (1) a dynamic mechanical subsystem, which for the purposes of this discussion includes a single-link robot manipulator and the motor rotor; (2) a dynamic electrical subsystem which includes all of the motor s relevant electrical e ects. d =x dt dx 1 = {[( L d dt J

(1)

L q ) Id + u

]Iq -

T sin

(2) (3) (4)

dId 1 = ( ud + x L q Iq - RId ) dt Ld dIq 1 = (uq - x L d Id dt Lq

RIq - x u f )

where (1) and (2) present the dynamics of mechanical subsystem, and (3) and (4) are the dynamic electrical subsystem. In these equations, uq and ud are the input control voltages, Id and Iq are the motor armature current, R is the stator resistance, L d and L q are the self-inductances, J is the inertia angular momentum, and u f is the ux due to permanent magnet. For the above electromechanical model, we assume that the true states (i.e. , x , Id and Iq ) are all measurable. This model is obtained by using circuits theory principles and a particular dq reference frame. The control objective is to develop a link position tracking controller for the electromechanical dynamics of (1) (4) despite parametric uncertainty. In the paper we assume that all the motor parameters are unknown.
Remark 1 : It should be noted that it is not pompous to refer to the above `pendulum model as a single link direct drive robot. In this paper, the single-link robot manipulator is regarded as a general load for motor. The main purpose of this paper is not to deal with the di erent load for the permanent magnet synchronous motor, but rather to develop a strictly globally stable adaptive robust controller and demonstrate the usefulness of the backstepping techniques for permanent magnet synchronous motors. Actually, the proposed method is applicable to any load which can be expressed by unknown parameters and known function of states, for instance the velocity tracking control problem in which the load can be expressed as 2 a + bx + cx , where a, b, and c are unknown parameters. Remark 2 : In most existing control schemes for the PM synchronous motors, the controllers are designed based on the following reduced model

896

J.-X. Xu et al. d =x dt dx 1 = ( u f Iq - T sin ) dt J dIq 1 = (uq - RIq - x u f ) dt Lq

(5) (6) (7)

In this reduced model, component ud is adjusted to regulate current state Id and it is supposed that Id exactly equals zero. Based on this reduced model, the position or velocity control is only directly related to the command voltage uq . Obviously, based on the reduced model, the control design will result in only a locally stable controller.
Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

3.

Adaptive backstepping control

In this section, the adaptive backstepping technique is used to develop a strictly globally stable controller for permanent magnet synchronous motors under the assumption that all the unknown system parameters are constant. The design procedures are presented in detail as follows.
ref 3.1. Design of auxiliary reference current Iq For the given desired tracking state d (t) , de ne a quantity s

where d (t) is at least twice continuously di erentiable. Di erentiating (8), multiplying by J and substituting the mechanical subsystem dynamics of (2), yields Js Divide (9) by u where
f

= ce + e,

e = -

to be

c>0

(8)

1 = Jce + [( L d Js

L q ) Id + u
>

]Iq -

T sin - Jx

(9) (10)

and rearrange terms to yield are de ned as

and

1=a

1u 1

+ Iq + L Id Iq

u 1

>

J , T and L are de ned as

= J
J
u

[ , - T ],
, f
ref

u 1

>

= ce - x
L =

, sin ] u
f

(11)

J =

T =

T
u

, f

Ld

Lq

De ne an auxiliary reference current Iq =

ref Iq
>

as k1 s
1

- a ^1 u 1 -

where k1 is a positive constant gain. The parameter estimate a ^1 = a ^11 , a ^12 generated by the following adaptive law

]
>

(12)
is

G 1

2 2

a ^1 = G
is a positive de nite matrix.

1u 1 s 1

(13)

Remark 3 . In the backstepping techniques, embedded control variable is an important concept. In the parametric-pure feedback form, the embedded control signal is designed as a full feedforward compensator which would provide good tracking performance. But in our problem, due to the nonlinear term L Id Iq shown in (10), the

Control schemes for PM synchronous motors

897

embedded control can not be designed as a full feedforward compensator. Thus the ref named auxiliary reference current Iq is designed as a partial compensator. The nonlinear term L Id Iq will be treated later by using adaptive techniques.

3.2. Design of control voltage ud In the adaptive backstepping method, the control voltage ud is designed not only to regulate the current Id , but also to compensate the nonlinear term L Id Iq in (10). To design ud , s 2 is selected as s 2 = Id . Di erentiating s 2 with respect to time, multiplying by L d and substituting the electrical subsystem dynamics (3), we have

L ds
Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

where

=a

2u 2

>

+ ud
>

(14)

and

u 2

are given by

>

[- R, L ],
q
>

u 2

= Id ,x Iq

(15) (16) (17) (18)

The control voltage ud and the corresponding adaptive law are chosen to be ud = - a ^ 2 u
2

a ^2 = C

- L^ s

1 Iq

k2 s

2u 2 s 2

L = g s 1 s 2 Iq where k2 and g are positive constant gains. C

2

2 2

is a positive de nite matrix.

Remark 4 : Usually the control voltage ud is designed to assure Id vanish. In the proposed method, ud is used to handle the nonlinear term Id Iq while ensuring that Id tends to zero, which will be shown later. This causes non-zero of Id in the adaptation period but the global stability of the closed-loop system is guaranteed from the beginning.

3.3. Design of control voltage uq The last step of the proposed design procedure is to choose control voltage uq to ref ensure that the actual current Iq approaches Iq . De ne s 3 as
s
3 ref = Iq - Iq

(19) (20)

Di erentiating (19), multiplying by L q and substituting (4), we obtain L qs where the unknown constant parameter vector a vector u 4 R 7 are de ned (see Appendix A) as

= uq + a

4u 4 4

>

and the known regression

4
>

>

[[

L d,

- R, - u f , ,

Lq

L , Lq J , - Lq T , Lq J J
f

]
]

u 4

= x Id , Id , x

(k1 + ca ^11 ) Iq , (k 1 + ca ^11 ) Id Iq ,

1u 1
>

(k1 + ca ^ 11 ) sin , k1 (ce - x d ) + s

- 1u
1

+ Ax

Design the control voltage uq with the corresponding adaptive law as

898

J.-X. Xu et al. uq =
> - a ^4 u 4 -

k3 s
7 7

(21) (22)

where k3 is a positive constant gain, C 4 R is a positive de nite matrix. The proposed adaptive backstepping controller ensures good tracking performance for the closed-loop electromechanical system dynamics as described by the following theorem. The proposed adaptive controller ensures global asymptotic position tracking for the permanent magnet synchronous motor (1) (4), as shown
Theorem 1:

a ^4 = G

4u 4 s 3

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

lim e = 0 lim e = 0 t
t

(23) (24)

Proof:

See Appendix B. As a summary, the control algorithm is given below.

ref

h
is

Step 1. Calculating s 1 and vector u 1 ; then the auxiliary reference current Iq generated together with adaptation of a 1 as follows
ref Iq = - a ^ 1 u
>

k1 s

(25) (26)

a ^1 = G

1 u 1s 1

Step 2. Calculating u 2 ; then control law for the voltage ud and the adaptive mechanisms for a 2 and L are given as ud =

a ^2 = G

> - a ^2 u 2 - L^ s

1 Iq

k2s

(27) (28) (29)

2u 2 s 2

L = g s 1 s 2 Iq

Step 3. Calculating s 3 , Ax , and vector u 4 ; then the control law for voltage uq and the associated adaptation law are designed as uq = - a ^ 4 u
>

k3s

(30) (31)

a ^4 = G

4 u 4s 3

Remark 5: In the above design procedure, the position tracking controller is developed by de ning s 1 = ce + e. It is easy to extend the proposed control procedure to velocity tracking control problems by de ning s 1 = e. The corresponding function vectors u 1 , u 2 , and u 4 are modi ed with c = 0.

4.

Robust backstepping control

In adaptive control, the unknown parameters are considered to be constant. Unfortunately, some parameters of PM synchronous motors may be time varying within certain known bounds, for instance the stator resistor R. In this section, taking into account the variations of system parameters, a robust position controller for the PM synchronous motor dynamics (1) (4) are developed. The pre-requirement

Control schemes for PM synchronous motors

899

for the robust control method is that bounds for all the parametric quantities in the electromechanical system are available. In the subsequent derivation procedure, unless otherwise speci ed, the notations are de ned the same as that for the adaptive backstepping controller. First, the ref auxiliary reference current Iq is selected as Iq = - a
ref 1u 1
>

k1 s

vI

(32)

where a 1 is a constant parameter vector which represents a `best-guess for the unknown parameter vector a 1 de ned in (11). vI is an auxiliary robust controller used to compensate for the mismatching between a 1 and a 1 :
Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

vI =

i s 1i

s 1q

2 I

m q Im

+e

I I

(33)
and q
Im

where e I > 0 is a constant design parameter, and q functions de ned as

q
I

are positive scalar

=i =

u 1

i a ~ 1b,
q

Im

=i
m

u 1 m ~ 1b

i a

(34)

where i i denotes the Euclidean norm, and i i

is a norm de ned to be
R

i wi

w > w + d

d , for all w

(35) (36)

with d being a positive constant. a ~1b is a bounding constant selected according to

a ~1b > i a ~1 i = i a 1 - a 1 i

The second step is the design of the voltage control ud . In the robust controller, ud is designed to be ud =

-a

2u 2

>

- L s

1 Iq

k2 s

vd

(37)

where k2 is a positive constant gain. The auxiliary controller vd is designed as ~ (38) vd = ( L b |s 1 Iq | + q d ) sign s 2

where q

is a positive scalar function de ned as

q
d

=i

u 2

i a ~2b

(39) (40)

a ~2b > i a ~2 i = i a 2 - a 2 i
~ Similarly, L b is a bounding constant chosen to be ~ ~ L b > | L | = |L - L

(41)

The last step in the procedure is to design the voltage control uq . Similar to the adaptive control design procedure, di erentiating (19), multiplying by L q and substituting (4), we obtain L qs

where the unknown constant parameter vector a 7 vector u 5 R are de ned as (see Appendix C)

= uq + a

5u 5 5

>

(42)

and the known regression

900

J.-X. Xu et al.

5
>

>

[[

L d,

- R, - u f , ,

Lq

J
11

Lq

L , J

- Lq T , J

Lq

]
+ H3 ) Id Iq ,

u 5

= x Id , Id , x
( k1 + ca
11

( k 1 + ca

+ H3) Iq , (k1 + ca

11

+ H3 ) sin , k1 (ce - x d ) + Ax + H4
5u 5
>

Now design the robust voltage control uq to be uq = - a

k3 s

vq

(43) (44) (45)

where k3 is a positive constant gain. The auxiliary controller vq is designed as vq = q

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

q sign s 3

where q

is a positive scalar function de ned as the following q q = i u 5 i a ~ 5b , a ~5b > i a ~ 5i = i a 5 - a 5 i

The proposed robust voltage controller ensures good position tracking performance for the electromechanical system dynamics (1) (4) as described by the following theorem. The proposed robust controller ensures globally bounded position tracking for the permanent magnet synchronous motor (1) (4). The tracking trajectory will enter the set
Theorem 2:

s D={

, s 2, s

3 :

2 1

+s

2 2

+s

2 3

< k e

}
(46)

and remain in it, where k is a positive constant

k =

max { , L d , L q } J . 2 min { 1 , k2 , k3 }min { , L d , L q } k J

Proof:

See Appendix D.

Note that the bound of the residual set can be made arbitrarily small by choosing a smaller e I at the expense of increasing the control authority in vI .
Remark 6:

5.

Adaptive robust backstepping control using switching function

In 3 and 4, adaptive and robust control methods for permanent magnet synchronous motors have been developed. We notice that, in the robust control scheme, the voltage control signal uq tends to be excessively large due to the terms which contain the derivative of the auxiliary controller vI used in the auxiliary reference ref current Iq , and is inevitable in the robust backstepping control scheme. Besides, the robust backstepping control scheme becomes very complicated in comparison with the adaptive control scheme. In practical applications, the parameters J and T are constants but may vary in a wide range due to the variation of payload. On the other hand, the unknown motor parameters have much less deviations from its rated values (nominal values) in comparison with that of load. Therefore, it would be more appropriate for us to deal with the unknown parameters J and T by using adaptive techniques and treating the bounded motor parameters by using robust methods. In this way, we

Control schemes for PM synchronous motors

901

can greatly simplify the controller design and reduce the control authority because ref the vI term is removed from the auxiliary reference current Iq . ref In the adaptive robust control method, auxiliary reference current Iq is designed using the adaptive method given by (12) and (13) Iq =
ref
> - a ^1 u 1 -

k1s

(47) (48)

a ^1 = G

1u 1 s 1

1 is chosen the same as previous. ud is designed the same as the robust control voltage given by (37) (41), namely

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

- L s 1 Iq - k2 s 2 ~ vds = ( L b |s 1 Iq | + q d ) sign s 2
ud =
2u 2
>

-a

vds

(49) (50)

Based on the relation (20) L qs uq =

>

= uq + a

4u 4

>

(51)

the control voltage uq now is designed as

-a

4u 4

k3 s

s
3

vqs

(52) (53)

where k3 is a positive constant gain. The auxiliary controller vqs is designed as vqs = q where q
q q

sign s

is a positive scalar function de ned as follows q q = i u 4 i a ~ 4b , a ~4b > i a ~4 i = i a 4 -

a 4i

(54)

For the adaptive robust controller (47) (54), we have the following theorem. The proposed adaptive robust control law (47) (54) ensures global asymptotic position tracking for the permanent magnet synchronous motor (1) (4), i.e.
Theorem 3:
t t

lim e = 0 lim e = 0

(55) (56)

Proof:

See Appendix E.

It is worthwhile pointing out that, owing to the absence of the vI ref term in the auxiliary reference current Iq , the required control uq now is much simpler than that of the robust one. In the robust control scheme, uq is a function of u 5 which includes terms such as x dx d . In the adaptive robust scheme, uq is a function of u 4 which only contains terms such as x d . Hence much lower control pro les can be expected, especially for the cases where fast system responses are required.
Remark 7:

Remark 8: Although the above controller guarantees the global stability of the system (1) (4), problems may be caused by the incorporation of switching functions. On the switching surface there might be some problems related to existence

902

J.-X. Xu et al.

and uniqueness of solutions. From the practical point of view, discontinuous functions also give rise to implementation problems and may excite high frequency modes of unmodelled dynamics. Hence our objective in the next section is to replace the discontinuous switching functions by continuous functions.

6.

Continuo us adaptive robust backstepping control with -modi cation scheme

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

In this section, continuous functions are used in the adaptive robust scheme to replace the signum functions. A new update scheme, referred to as the -modi cation scheme, is proposed to cease parameter adaptation in accordance with the adaptive robust control law. The novel property possessed by this new adaptive robust backstepping control scheme is that, it guarantees the uniform boundedness of the system and at the same time, is able to allow the tracking precision to be arbitrarily designated through the choice of suitable control design parameters. ref First, the auxiliary reference current Iq is designed as Iq =
> - a ^1 u 1 - k1 s 1 a ^1 = G 1 u 1 s 1 - G 1 a ^1

ref

(57) (58)
e

where

is chosen the same as previous and

0 ks 1 ( e
0

- |s 1 |)

|s 1 |

is de ned as
0

otherwise

(59)

where ks 1 is a positive constant and e 0 is a small positive value determined by the desired tracking error bound. If the tracking error bound is speci ed by

|e| = | - d |
e
0

emin ,

then

= cemin
1 Iq

The robust control voltage ud is designed to be ud = - a

2u 2
>

where k2 is a positive constant gain. The auxiliary controller vds is designed as 2 2~2 s 2 s 1 Iq L b s 2q 2 d (61) vds = + ~ i s 2i q d + e 2 i s 1s 2Iq i L b + e 1

- L s

k2 s

vds

(60)

~ where q d and L b are de ned in (39) and (41) respectively. e design parameters. Based on the relation (20)

> 0 and e

> 0 are

L qs uq =

>

= uq + a

4u 4

>

(62)

the control voltage uq now is designed as

-a

4u 4

k3 s

vqs

(63)

where k3 is a positive constant gain. The auxiliary controller vqs is designed to be

Control schemes for PM synchronous motors vqs =

903

i s 3i

s 3q
q
q

2 q

+e

(64)

where e 3 > 0 is a design parameter. For the above adaptive robust controller, we have the following theorem. By choosing the control gain k1 (e + c1 ) /e 2 with e = e 1 + e 2 + e 3 0 and c1 > 0, the proposed continuous adaptive robust control law (57) (64) ensures that the system trajectory as well as the parameter estimation error will enter the set > > s D = { 1 , s 2 , s 3 , a~ 1 : |s 1 | e 0 ; s 2 + s 2 + s 2 + a ~1 a ~1 < k (ks e 0 a 1 a 1 + e ) } (65) 1 2 3
Theorem 4:

in a nite time, where k is de ned to be

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

k =

1 max { , L d , L q , (C - ) } J 1 2 min { 0 , k1 , k2 , k3 }min { , L d , L q , (C - 1 ) } k J 1

(A) and (A) indicate the maximum and minimum eigenvalues of the matrix A

respectively. k0 > 0 is a constant to be explained later.

Proof:

The following positive de nite function V is selected > 2 2 2 1 V = 1 J s 1 + 1 L d s 2 + 1 L q s 3 + 1 a ~1 G - a ~1 1

2

where

(66)

a~ 1 = a 1 - a ^1
Taking the derivative of V along the trajectory of the overall electromechanical dynamics (1) (4) with the adaptive robust control law (57) (64), we have > V = J s 1s 1 + L d s 2s 2 + L qs 3s 3 + a ~1 a ~1

= - k1 s
+ s 3 (a

2 1
>

ref + s 1 (Iq - Iq ) + L s 1 Id Iq + s 2 (a 4u 4

-a

4u 4

>

>

2u 2
>

>

-a
>

2u 2

>

k3 s

- L s
s 2 vds

1 Iq

k2 s

vds )

vqs ) + a ~1 a ^1
2

= - k1 s

2 1

+ s 1 (L

- L ) Id Iq -

k2 s

2 2

+ s 2 (a

-a

2) u 2

> - k3s 2 + s 3 (a 4 - a 4) u 4 - s 3vqs + a ~ 1 a ^1 3 ~ - k1 s 2 - k2 s 2 - k3 s 2 + L b |s 1Id Iq | + q d |s 2 | + q q|s 3| 1 2 3

s 2 vds

s 3 vqs - a ~ 1 a ^1
>

- k1 s 2 1 - k1 s 2 1 - k1 s 2 1 - k1 s 2 1
where

k2 s k2 s k2 s k2 s

2 2 2 2

k3 s k3 s k3 s k3 s

2 3 2 3 2 3 2 3

+e

+e

2
>

+e

3 1

+ a ~ 1 a ^1
>

2 22 2

+ e + a ~ 1 (a +e

- a~1)
1 1
>

> > - a ~1 a ~1 + a ~ 1 a

1 ~> 2 1

a a ~1 + 1 a 1 a 2
e

+e

(67)

=e 1+e 2+e

904 From (59), if |s

1

J.-X. Xu et al.

e 0 , we have

V If we choose k1 to be

- k1 s 2 1
k1

k2 s

2 2

k3 s

2 3

+e

(68)

e + c1 e
2 0

where c1 is an arbitrary small positive constant, then we can easily get V Notice that V is a continuous function, so obviously there exists a constant e that:
Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

- c1

" |s

|>e

(68)

0 such
(69)

0<e

V <0 From (67), we can see that V < 0 if: k1 s

2 1

0 < e 0
> 1 a 2

" |s
1
>

| > e 0

a 1+e

Note that a 1 and e are constants and is a function of s 1 in terms of the de nition (59). Hence e 0 can be easily determined by solving the equation

k1 s
2

2 1

= 1 a 2

>

a 1+e
1 0) a

Substituting in terms of (59) and replacing s k1 e

by e 0 , we get:
1a 1+ e
>

The positive solution of equation (70) is the desired e

e

= 1 ks 1 ( e 2

-
e

(70)

> - ks a 1 a
1

> 2 > (ks 1 a 1 a 1 ) + 4k1 (ks 1 a 1 a 1 e 0 + 2e )

0 :

2k1

As a consequence, when |s 1 | > e 0 , the system will enter |s the other hand, when |s 1 | e 0 , it is obvious that ks 1 ( e V

| > e 0 in a

nite time. On

From (59), (67) and (71) we obtain

- e 0)
>

ks 1 e

(71)

- k V + 1 a 1 a 1 + e 2 > - k V + ks e 0 a 1 a 1 + e
1

" |s

| e 0

(72)

where k= 2 min { 0 , k 1 , k 2 , k 3 } k max { , L d , L q , (G - 1 ) } J 1

k0 =

1 2 ks

(e

By integrating (72) we can establish that

- e 0) > 0

(73)

Control schemes for PM synchronous motors V ( t)

905

1 > kt (73) e- V (t = 0) + (ks 1 e 0 a 1 a 1 + e ) k which implies that s 1 , s 2 , s 3 and a ~1 converge exponentially to the residual set > > 2 2 2 (74) s D = { 1 , s 2 , s 3 , a ~ 1 : s 1 + s 2 + s 3 + a ~ 1 a ~ 1 < k ( ks 1 e 0 a 1 a 1 + e ) }

where k =

1 max { , L d , L q , (G - ) } J 1 2 min { 0 , k1 , k2 , k3 }min { , L d , L q , (G - 1 ) } k J 1

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Finally, from (68), (69) and (74) we can conclude that s 1 , s 2 , s exponentially to the residual set > 2 2 2 s D = { 1 , s 2 , s 3 , a ~1 : |s 1 | e 0 ; s 1 + s 2 + s 3 + a ~1 a ~1 < k (ks e

, and a ~1 converge
h

0 1

> a a 1 + e )}

Remark 9: Analogous to the selection of the gain k1 , we can choose gains k2 and k3 such that the tracking error bounds of the quantities s 2 and s 3 can be speci ed through the design. Remark 10: In Liao et al. (1990 ) and Brogliato and Neto (1995), the synthesis of the adaptive and robust control schemes with speci ed tracking error bounds were presented. However, those methods are valid only for one-step design of the systems. In other words, their algorithms cannot be extended to the backstepping design which requires that the input be di erentiable at each step and consists of at least two steps. It should be noted that, the control design for the PM synchronous motor is essentially two steps: the ctitious auxiliary reference current input and the actual voltage inputs ud and uq ; thus the method proposed here is applicable.

7.

S imulation studies

The permanent magnet synchronous motor with following parameter is used to demonstrate the control performance 3 3 3 2 L d = 25. 0 10- H, L q = 30. 0 10- H, J = 1. 625 10- Kg m ,
u
f

= 0.90 N m /A,

R = 5. 0 V

T = 2. 2816 Kg A m /N s2

>

= 0.0018,- 2. 535
2

The following control parameters

= 1.2I2

c = 5. 0 k3 = 2. 0

are used for all the proposed control schemes. The estimates a 1 , a 2 and a 4 are chosen 15% larger than their real values, while a ~2b , a ~4b and a ~5b are chosen based on the assumption that each of the electromechanical parameters in (2) and (5) would be o 20% from the estimates a 2 , a 4 and a 5 . The desired trajectory is
d

k1 = 8. 0,

k 2 = 2. 0,

3 p ( p 1 - e- 0. 1t ) sin t

(5)

906

J.-X. Xu et al.

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 1.

(a) Evolution of input voltage uq ; (b) Evolution of input voltage ud .

The initial states are (0) = 0. 1 rad, x (0) = 0. 1 rad /s, Id (0) = 0. 05 A, Iq (0) = 0. 01 A. The initial values of all the parameter estimates are set zero. 7.1. Adaptive backstepping control The control law (25) (31) is found to yield a `good position tracking performance with the following control parameters
C 2

= 0.05I2

g = 0.05,

C 4

= 0.05I7

The control inputs ud and uq are given in gure 1. The tracking error is shown in gure 2. Figure 3(a) shows the corresponding armature current Iq compared with ref auxiliary reference current Iq while Id is shown in gure 3(b). It is easy to see that ref the current Iq asymptotically converges to Iq . During the adaptation period, ud is used to counteract the e ect of the nonlinear term Id Iq in the mechanical subsystem. 7.2. Robust backstepping control Taking into account the variations of system parameters, the robust control method is used with the following control parameters e I = 0. 1, a ~1b = 0.5, a ~2b = 1.0
~ L b = 0. 002,

The simulation results are shown in gures 4 6. It is obvious that robust control method is found to yield the better position tracking performance compared with the

a ~5b = 3.0

Control schemes for PM synchronous motors

907

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 2. (a) Position tracking error e; (b) Velocity tracking error e.

ref Figure 3. (a) Evolution of current Iq (solid-line) and Iq (dot-line ); (b) Evolution of current Id .

908

J.-X. Xu et al.

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 4.

(a) Evolution of input voltage uq ; (b) Evolution of input voltage ud .

Figure 5.

(a) Position tracking error e; (b) Velocity tracking error e.

Control schemes for PM synchronous motors

909

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 6.

( a): Evolution of current Iq ; (b) Evolution of current Id .

adaptive control method, as indicated in gure 5. However, the control voltage uq , shown in gure 4(a), is much higher than that of adaptive control method due to the terms which contain the derivative of the auxiliary controller vI . 7.3. Adaptive robust backstepping control method using switching function The additional control parameter

a ~4b = 3.0
is selected in this control scheme. The parameter estimate a ^1 is generated by the adaptation law (48). Simulation results are depicted in gures 7 9. We can see that the control scheme achieves good tracking performance ( gure 8), while the control voltage uq is greatly reduced compared with the robust backstepping control signal in gure 4(a). Due to the incorporation of switching functions, high frequency chattering phenomenon can still be observed in the control pro les. 7.4. Continuous adaptive robust backstepping control method with -modi cation scheme The following control parameters are chosen in the adaptive robust control scheme proposed in 6:
e
1

=e

=e

= 2. 0,

ks 1 = 8

910

J.-X. Xu et al.

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 7.

(a) Evolution of input voltage ud ; (b) Evolution of input voltage uq .

Figure 8.

(a) Position tracking error e; (b) Velocity tracking error e.

Control schemes for PM synchronous motors

911

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 9.

(a) Evolution of current Id ; (b) Evolution of current Iq .

The parameter estimate a ^ 1 is generated by the adaptation law (48). The simulation results are shown in gures 10 13. We can see that the tracking performance ( gure 11) is similar to that of 7.3, but the chattering phenomenon has been reduced signi cantly, hence the control signals are much smoother (as shown in gures 10 and 12). The parameter estimate a ^ 1 is shown in gure 13.
Remark 11: The control `chattering is highly dependent on the choice of the design parameters e 1 , e 2 and e 3 . The greater the three parameters, the smaller the `chattering, but the greater the size of the residual set D (65). Notice that if the parameters are set to zero, the control scheme in 6 is exactly the same as that in 5. Remark 12: Comparing the four control schemes developed in this paper, it is clear that the continuous adaptive robust control scheme is the best in the sense of making `balance in terms of the tracking accuracy, control algorithm complexity, control signal authority and smoothness, and design degrees of freedom. 8. Conclusions

In this paper, four control schemes are developed for a permanent magnet synchronous motor which is not exactly in the parametric-pure feedback form. The adaptive backstepping control is rst proposed to achieve strictly globally asymptotic convergence of the system tracking error. Taking into account parameter variations in electromechanical system, the robust backstepping control scheme is

912

J.-X. Xu et al.

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 10.

(a) Evolution of input voltage ud ; (b) Evolution of input voltage uq .

Figure 11.

(a) Position tracking error e; (b) Velocity tracking error e.

Control schemes for PM synchronous motors

913

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Figure 12.

(a) Evolution of current Id ; (b) Evolution of current Iq .

Figure 13.

(a) Evolution of parameter a

11 ;

(b) Evolution of parameter a

12 .

914

J.-X. Xu et al.

proposed which achieves better tracking performance at the expense of larger control authorities. To obtain better tracking control performance and yet retain low control pro les, the adaptive and robust backstepping control scheme using switching functions are synthesized. In the adaptive robust controller, the robust part is used to suppress the variations of system parameters with upper bounds while the unpredictable load is estimated by the adaptive part. In order to smooth the control input signals, we further develop the continuous adaptive robust backstepping control method associated with the -modi cation scheme, which achieves signi cant reduction of the control chattering and at the same time guarantees the uniform boundedness of the system tracking error with speci ed accuracy. Simulation studies con rm the validity of the four proposed control schemes.
Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Appendix A: De nitions of

and }

Di erentiating (19), yields

s

3 = Iq - Iq

ref

(75)

From (12), we have Iq =

ref - (s
=-s

1C 1u 1) u 1
>

>

> - a ^1 u 1 - k1 s

1u 1

C 1u 1

> - a ^1 u 1 -

k1 (a J

> For a ^ 1 u >

c a ^1 u 1 = a ^11 [ (x - x d ) - x d ]+ a ^12x cos

1,

it follows

1u 1

>

+ Iq + L Id Iq )

(76)

= =a
where

ca ^ 11 [( L d - L q ) Id Iq + u f Iq - T sin ]- a ^11 (cx J

+ x d ) + a ^12x cos

3u 3

>

+ Ax

(77)

a 3, u

and Ax are de ned by

3
>

>

Ld

Lq u

, Jf ,-

T J

(78) (79) (80)

u 3

= ca ^ 11 Id Iq , Iq , sin

Ax =

- a ^11 (cx
RId
1

+ x d ) + a ^ 12x cos

Multiplying (75) by L q and substituting (4), we obtain L qs

3 = uq - x

L d Id
1u 1
> >

G
1

-x
u
>

+ Lq s

- 1u

+a

3u 3

+ Ax +

k1 (a J

= uq + a
where

1u 1

>

+ Iq + L Id Iq )

]
(81)

4u 4

and

u 4

are de ned as

Control schemes for PM synchronous motors

915

> >

u 4

L - R, - u f , L q Jf , L q J , - L q T , L q J x = [ Id , Id , x , (k1 + ca ^11 ) Iq , (k 1 + ca ^11 ) Id Iq , (k1 + ca ^ 11 ) sin , > 1 k1 (ce - x d ) + s 1 u 1 C - u 1 + Ax ] 1

[-

L d,

Append ix B:

Proof of Theorem 1

First, de ne the following positive de nite functions V 1 and V 2 > 2 1 V 1 = 1 J s 1 + 1 ( a 1 - a ^1 ) C - ( a 1 - a ^ 1 ) 1

2

(82) (83)

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

V2 = 1 L ds 2 It follows that V 1 = J s 1s

2 2

+ 1 (a 2 - 1a ^
1

> - a ^2 )

C 2

(a

- a ^2 )

11u 1 1u 1 1 2 1
> >

(a

- a ^1 ) > G

= s 1 (a = s 1 (a =s
1

+ Iq + L Id Iq )

> - ( a 1 - a ^1 ) G

- 1a ^
1

1
> - a ^1 ) G

ref ref + Iq + Iq - Iq + L Id Iq ) - (a

- 1a ^
1 1

1 1 C - a ^1 1

[ (a

> - a ^1 )

u 1

k1 s

ref + Iq - Iq + L Id Iq

] - (a

> - a ^1 )

= - k1s
and

ref + s 1 (Iq - Iq ) + L s 1 Id Iq

(84)

V 2 = L d s 2s

=s

[(a - a ^ )
2 2
>

22 2

(a

> - a ^2) G

- 1a ^
2

u 2

- L^ s

1 s 2 Iq

k2 s

]-

(a

> - a ^2)

1 C - a ^2 2

= - k2 s

- L^ s
q 3

1 s 2 Iq

(85)

To complete the analysis of the adaptive backstepping controller, the following positive de nite function V 3 is selected > 2 (86) V = V + V + 1 L s + 1 (a a ^ ) C - 1 (a a ^ ) + 1 g - 1 ( L L^ ) 2
3 1 2 2 2 4

Taking the derivative of V 3 along the trajectory of the overall electromechanical dynamics (1) (4) and substituting V 1 , V 2 yields

V3 = V 1 + V 2 + L qs 3s

33

(a

> - a ^4 )

C 4

a ^ 4 - g - 1 ( L - L^ ) L^
2 2

= - k1 s

2 1

ref + s 1 (Iq - Iq ) + L s 1 Id Iq - k2 s
>

+ s 3 (- a ^ 4 u

2 2 2 2

k3 s k3 s k3 s

s
2 3 2 3

+a

4u

>

- L^ s 1 s 2 Iq ^ > -1^ 4 ) - (a 4 - a 4 ) G 4 a 4 ^ - L ) - s

g - 1 ( L
3s

= - k1 s = - k1 s

2 1 2 1

k2 s k2 s

ref + s 1 (Iq - Iq ) + s 1 s 2 Iq ( L

- L^ ) L^ ^ ^ - 1 (L - L ) L 1- g
(87)

916

J.-X. Xu et al.

This fact implies that s 1 , Id ( s 2 ), s 3 , and the parameter estimates are bounded and square integrable. Consequently, it is easy to verify that s 1 , s 2 and s 3 are bounded. These facts, together with Barbalats Lemma, allow us to conclude that

t t t

lim s lim s lim s

1 2 3

=0 =0 =0

From the de nition of s 1 , it shows that the control task has been achieved by this controller and the control system is global stable, i.e.
Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

t t

lim e = 0 lim e = 0

Appendix C:

De nitions of

and }

Di erentiating (19), yields

s

From (32), we have

3 = Iq - Iq
1u 1
>

ref

(88) (89)

ref Iq = - a

For vI , it follows that vI =

- k 1 s 1 - vI
s 1q
2 I 2 + e I) Im

s 1q

i s 1i

2 I

+ 2s

~ 1 u 1 u 1 1b

m q Im 2 I

>

+e

(i s 1 i m q

(i s 1 i

m q Im

+ i s 1i

m q Im )

i s 1i

s 1q s 1q
1

m q Im

+e
2 I

> 2s 1 u 1 u 1 a ~ 1b i s 1 i m q Im + e I

u > u 1 + d

i s 1i
>

) (i s
q
2 I

~ m u 1 u 1 1b

>

1 m q Im

+ e I)

(
s

2
1

s 2q 2 s 1q 1 I

+d

) (i s

Im 2 + e I)

1 m q Im

= H1 s 1 + H2 u
H1 =

1u 1

(90)
s 1q I q
2 2

where H1 and H2 are given by

i
s

mq

Im

+e

I I

( (

Im m q Im 2 + e I)

+d

H2 =

2s 1 a ~1b i s 1 i m q Im + e
1

i s 1i m a ~1b > 2 u 1 u 1 + d (i s 1i m q Im + e I )
s 1q
2 I

(i s 1 i

(91)

In terms of de nitions of s

in (8) and

u 1

in (11), we have

(92)

Control schemes for PM synchronous motors vI = H1 ce + x

1

917

[ - x ]+ H [(c e - cx ) (x - x ) + x x = [ + H (c e - cx ) ] + H (ce - x ) H x + H [ sin cos + x x - (c e - cx )x - cex ] H = I { ( L - L ) I + u ] - T sin }+ H J [

d 2 2 d d 2 2 d 1 d 2 2 d d d d d 3 d q d f q 4 2 H3 = H1 + H2 (c e - cx

sin cos + x

dx

d - cx d e

(93)

where H3 and H4 are given by

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

H4 = H1 (ce - x

x d ) + H2 [ sin cos + x

d) dx

(94) d

(c2 e - cx d )x

- cex d ]

(95)

Similar to the derivation of (77), we get

1u 1

>

=a = =a
ca

11

[c(x - x ) - x ]+ a
d d

12x

cos

11

J
>

[( L d + Ax

L q ) Id Iq + u f Iq - T sin ]-

11 (cx d

+ x d ) + a

12x

cos

3u 3

(96)

where
u 3
>

= ca

11

[I I , I , sin ]
d q q

(97)
12x

Ax = - a

11 (cx d

+ x d) + a

cos

(98)

Using (93) and (96), Iq is given by Iq = - a

ref

ref

1u 1
>

>

- k1 s 1 - vI
Ax

=-a

3u 3

k1 (a J

H3 {(L d J

[ - L )I
q f

1u 1

>

+ Iq + L Id Iq )

+u

]I - T sin }- H
q

(99)

Multiplying (88) by L q and substituting (4) and (99), we obtain L qs

= uq - x L d Id - RId - x u
+
H3 {(L d J
5u 5
>

+ Lq

[ - L )I
q

+u

]I q

3u 3

>

+ Ax +

T sin }+ H4

= uq + a
where

k1 (a J

1u 1

>

+ Iq + L Id Iq )

(100)

and

u 5

are de ned as

918

J.-X. Xu et al.
u L [- L , - R, - u , L J , L J , - L T , L ] J x = [ I , I , x , (k + ca + H ) I , (k + ca + H ) I I , (k k (ce - x ) + A + H ]

5
>

>

u 5

11

11

d q

+ ca

11

+ H3 ) sin ,

Appendix D:

Proof of Theorem 2

The following Lyapunov function candidate V is selected V = 1J s 2

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Taking the derivative of V along the trajectory of the overall electromechanical dynamics (1) (4), substituting (10), (14), (100) and the robust controller (32), (37) and (43) we have V = J s 1s

2 1

+ 1 Lds 2

2 2

+ 1 L qs 2

2 3

(101)

>

+ L d s 2s
>

+ L qs 3s

= s 1 (a

1u 1

ref ref + Iq + Iq - Iq + L Id Iq ) + s 2 (a

+ s 3 (a

5u 5 2 1

-a
1 2 5 1 2 5

5u 5

>

> > >

k3 s

2u 2

>

-a

2u 2

>

- L s

1 Iq

k2 s

vd )

vq )

= - k1 s

+ s 1 (a + s 2 (a + s 3 (a + s 1 (a + s 2 (a + s 3 (a

-a -a -a -a -a -a
1

1) u 1 2) u 2 5) u
>

ref + s 1 ( Iq - Iq ) + s 1 L Id Iq - s 1 vI

2 k2 s 2

k3 s

2 3 2 1 2 2 2 3 2 1

- L s 1 s 2 Iq - s 5 - s 1 s 3 - s 3 vq
+ s 1 ( L

2 vd

= - k1 s
k2 s k3 s

1) u 1 2) u 2 5) u 5
> >

- L ) Id Iq -

s 1 vI

s 2 vd s 3 vq

- k1 s

> + s 1a ~1 u > + s 2 a ~2 u

k2 s

2 2

~ ~ - i s 1 i q 1 ImI + e I + s 1 L Id Iq - L b|s 1Id Iq | m 2 ~> 2 - q d |s 2 | - k3 s 3 + s 3 a 5 u 5 - q q |s 3 |

d

s q

2 2

(102) (103) (104)

In terms of the choices of q I , q V

and q q , we can obtain the following k2 s

2 2

- k1s 2 1
V

k3 s

2 3

+e

which is upper bounded by the expression

- kV + e

where k is de ned to be k= 2 min { 1 , k2 , k3 } k . max { , L d , L q } J

By integrating (104), we can establish that

(105)

Control schemes for PM synchronous motors V (t) which implies that s

1

919

e kt e- V (t = 0) +
3

,s

and s

I (1 - e- kt ) k converge exponentially to the residual set 3 :

(106)

s D={

, s 2, s

2 1

+s

2 2

+s

2 3

< k e

where k =

max { , L d , L q } J 1 1 = . k min { , L d , L q } 2 min { 1 , k2 , k3 }min { , L d , L q } J k J

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Append ix E:

Proof of Theorem 3

The following positive de nite function W is selected as > W = 1 J s 2 + 1 L s 2 + 1 L s 2 + 1 (a - a ) C - 1 (a

2

d 2

q 3

-a

1)

(107)

Taking the derivative of W along the trajectory of the overall electromechanical dynamics (1) (4) with adaptive robust control law (47) (54) , we have W = J s 1s

1 + L d s 2s 2 + L qs 3s 3 ref = - k 1 s 2 + s 1 (Iq - Iq ) + L s 1
+ s 3 (a
4u 4 2 1
>

1 Id Iq

+ s 2 (a
vqs )

2u 2

>

-a
>

2u 2

>

-a

4u 4

>

k3 s

- L s
s 2 vds

1 Iq

k2 s

vds )

= - k1s

+ s 1 (L

- L ) Id Iq - k2 s
>

2 2

+ s 2 (a

-a

2) u 2

>

- k3 s 2 + s 3 (a 4 - a 4) u 4 - s 3 vqs 3 ~ ~ = - k 1 s 2 - k2 s 2 - k3 s 2 + s 1 L Id Iq - L b |s 1 2 3 > - q d |s 2 | + s 3a ~ 5 u 5 - q q |s 3|
W

1 Id Iq

| + s 2 a ~2 u

(108)
2 2

Utilizing (34), (39) (41) , and (45), we can obtain

- k1 s 2 1
t t t

k2 s

k3 s

2 3

(109)

Similar to Theorem 1, the following results are concluded as lim s lim s lim s
1 2 3

=0 =0 =0

From the de nition of s 1 , it shows that

t t

lim e = 0 lim e = 0

920
References

Control schemes for PM synchronous motors

Downloaded By: [Canadian Research Knowledge Network] At: 12:41 3 August 2010

Barmish, B. R., Corless, M. J., and Leitmann, G., 1983, A new class of stabilizing controllers for uncertain dynamical systems. SIAM Journal of Control and Optimization, 21 , 246 255. Brogliato, B., and Trofino Neto, A., 1995, Practical stabilization of a class of nonlinear systems with partially known uncertainties. Automatica, 31, 145 150. Corless, M. J., and Leitmann, G., 1981. Continuous state feedback guaranteeing uniform ultimate boundedness for uncertin dynamic systems. IEEE Transactions on Automatic Control, 26 , 1139 1144. Dawson, D. M., Carroll , J. J., and Schneider, M., 1994. Integrator backstepping control of a brush dc motor turning a robotic load. IEEE Transactions on Control Systems Technology, 2 , 233 244. Fujimoto, Y., and Kawamura , A ., 1995, Robust servo-system based on two-edge-of-freedom control with sliding mode. IEEE Transactions on Industrial Electronics, 42 , 272 280. Gutman, S., 1979, Uncertain dynamical systems a lyapunov min max approach. IEEE Transactions on Automatic Control, 24 , 437 443. Ioannou, P., and Sun, J., 1996, Robust Adaptive Control. (New York: Prentice-Hall). Jiang , Z., and Pomet, J. B., 1995, Backstepping-based adaptive controller for uncertain nonhonolomic systems. Proceedings of the 34th CDC, pp. 1573 1578. Kanellakopoulos, I., Kokotovic, P. V., and Morse, A. S., 1991, Systematic design of adaptive controller for feedback linearizable systems. IEEE Transactions on Automatic Control, 11 , 1241 1253. Kokotovic , P., 1991, Foundations of Adaptive Control. (New York: Springer-Verlag, CCES). Liao, T.-L., Fu , L.-C. , and Hsu , C.-F., 1990, Adaptive robust tracking of nonlinear systems and with an application to a robotic manipulator. Systems & Control L etters, 15 , 339 348. Narendra , K., and A nnaswamy, A. M., 1989, Stable Adaptive Systems. (Englewood Cli s, NJ: Prentice-Hall). Qu , Z.,1993, Robust control of nonlinear uncertain systems under generalized matching conditions. Automatica, 29, 985 998. Sastry, S., and Isidori, A., 1989, Adaptive control of linearizable systems. IEEE Transactions on Automatic Control, 34 , 1123 1131. Slotine, J. J. E., and Li, W. P., 1992, Applied Nonlinear Control. (Englewood Cli s, NJ: Prentice-Hall). Taylor , D., Kokotovic, P. V., Marino, R. , and Kanellakopoulos, I., 1989, Adaptive regulation of nonlinear systems with unmodeled dynamics. IEEE Transactions on Automatic Control, 34, 405 412. Utkin, V. I., 1992, Sliding Modes in Control Optimization. (New York: Springer-Verlag, CCES). Xu , J.-X., Lee, T.-H., and Jia , Q.-W., An adaptive robust control scheme for a class of nonlinear uncertain systems. International Journal of Systems Science, 28 , 429 434.