This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Digital Object Identifier

A Non-smooth Composite Control

Approach for Direct Torque Control of
Permanent Magnet Synchronous
Machines
CHUNHUA XIA1 , SHENGHUI LI2 , YING SHI2 , XINGHUA ZHANG3 , ZHENXING SUN3 , AND
WENQING YIN1
1
School of Engineering, Nanjing Agricultural University, Nanjing, 210031 CHINA
2
School of Electronic and Optical Engineering, Nanjing University of Science and Technology ZiJin College, Nanjing, 210023 CHINA.
3
College of electric engineering and control science, Nanjing Tech University, Nanjing, 211816 CHINA.
Corresponding author: Wenqing Yin (e-mail:yinwq@njau.edu.cn).
This work is supported in part by Qing Lan Project (2016), in part by Natural Science Foundation of the Jiangsu Higher Education
Institutions of China (18KJB413005), in part by National Science Foundation of China (51477073) and in part by Basic Research Project
of Central Public Welfare Research Institutes (S201715).

ABSTRACT Considering the effects of disturbances in permanent magnet synchronous motor (PMSM),
in this paper, a non-smooth composite control approach, which includes finite time disturbance observer
(FTDO) for feedforward compensation and finite time control (FTC) for feedback control, is proposed
to improve the anti-disturbance performance of PMSM. First, the FTDO is used to estimate the lumped
disturbances, such as friction, parameter perturbation and load variation. Then, the observed value is added
into the speed controller as a feedforward compensation to eliminate the effect of disturbance. Second, FTC
is introduced into the feedback control design part. In the end, by utilizing Lyapunov theory, the stability
analysis of overall closed-loop system is demonstrated. In contrast to conventional asymptotically stable
control strategy, the proposed composite scheme can provide not only a faster dynamic response, but also a
stronger capacity of disturbance rejection. Simulation and experimental tests are presented to demonstrate
the superior properties of the proposed control scheme.

INDEX TERMS Permanent magnet synchronous machine (PMSM), direct torque control (DTC), distur-
bance rejection, finite-time disturbance observer (FTDO), finite-time control (FTC).

I. INTRODUCTION though it is simple, DTC is capable of generating fast torque

UE to the characteristics of low inertia, high efficiency,
D high power density, free maintenance, high torque-
current ratio and low noise [1], permanent magnet syn-
chronous motors (PMSMs) have been extensively used in
plenty of practical industrial applications, such as power
generations [2], autonomous mobile robots [3], aeronautics
and astronautics [4], electric vehicles [5], numerical con-
trol machine [6] and other industrial fields [7]. The high-
performance control system of PMSM often adopts two
typical control frameworks: Field orientated control (FOC)
and direct torque control (DTC), which are presented by
Blaschke in 1971 and M. Depenbrock in 1985, respectively
[8], [9]. Compared with FOC, DTC needs not coordinate
transformation and any current control. It controls the torque
and flux of PMSM according to the instantaneous errors. Al-
response. Moreover, since that DTC could minimise the use
of motor parameters, it is less susceptible to the changes of
model parameters. Thus, more and more researchers pay their
attention to DTC in recent years.
In actual operation, disturbances are an important factor
which will deteriorate the control performance of PMSM.
These disturbances may come from both in and out of the
PMSM systems, e.g. unmodeled dynamics, frictions and load
disturbances. The common approaches based on feedback
control, for example, proportional integral differential control
(PID), usually can not restrain the disturbances quickly,
in other words, these control schemes can suppress those
disturbances by means of feedback regulation in a relative-
ly slow way. With the developments of power converters
technology and high performance microprocessors, many

VOLUME 4, 2016 1
2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Author et al.: 11

advanced control methods (e.g. sliding model control (SMC) A. SYSTEM MODEL OF PMSM
[10], model predictive control (MPC) [11], adaptive fuzzy Under the rotor oriented coordinates d − q, the equations of
controller [12]) have been proposed to enhance the capacity a PMSM can be described as follow [30] :
of anti-disturbance of PMSM systems from different aspects. (
In addition to the control methods mentioned above, taking ψd = Ld id + ψr
(1)
a feed-forward compensation of disturbance into the design ψq = Lq iq
of controller can be considered as another effective and easy (
way to enhance the control performance of PMSM in this ud = Rs id + ψ̇d − ωψq
case. Because of the diversity of the disturbances in PMSM, (2)
uq = Rs iq + ψ̇q + ωψd
it is difficult to detect those disturbances by using instrument
3

and equipment, disturbance observation technology can be
 Te = np (ψd iq − ψq id )

seen as a good alternative. A variety of disturbance observers 2 (3)
have been presented, during the past few decades, for exam-  ω̇ = 1 Te + 1 TL + Bω

ple, disturbance observer (DOB) [13], sliding mode observer J J
(SMO) [14], equivalent input disturbance (EID) approach where ud and uq are d- and q- axis stator voltages, respective-
[15], perturbation observer [16], unknown input observer ly; id and iq are d- and q- axis stator current, respectively; Ld
(UIO) [17], extended state observer (ESO) [18] and finite and Lq are d- and q-axis stator inductances, respectively; ψd
time disturbance observer [19], [20]. and ψq are d- and q- axis stator flux linkage, respectively;
It is note worthy that for most of existing control methods, ω is the rotor angular velocity; B is the viscous frictional
the closed-loop system for PMSM is asymptotic stable, that coefficient; TL is the load torque; ψr is rotor flux linkage;
is to say, the system output is allowed to track asymptotically ψs is amplitudes of stator flux; Rs is the stator resistance; J
the reference signal. For PMSMs, it is crucial to improve is the moment of the total inertial; np is the number of pole
the system’s transient response [21]–[23]. In contrast to pairs.
asymptotically stable of closed-loop systems, finite-time sta- In the stationary stator reference frame, the model of
ble systems own some superiorities [24]–[26], for example, PMSM in the rotor reference frame can be expressed as
faster convergence rate at the equilibrium points, and stronger  Z
performance of anti-disturbance. Because of these merits 
 ψα = (uα − Riα )dt
mentioned above, it has been widely studied in engineering


 Z
and technology, for example, robotic manipulators [27], pow- (4)
ψ β = (uβ − Riβ )dt
er electronics [28] and serial elastic actuators [29].




Based on the above analysis, a non-smooth composite ψs = ψα2 + ψβ2

control method, which combines FTC and FTDO, is pro-
posed. The unknown lumped disturbances can be observed 3
np (ψα iβ − ψβ iα )
Te = (5)
by FTDO. By adding the observed value into the feedforward 2
path of controller, the capacity of disturbance rejection can be where uα and uβ are α- and β- axis stator voltages, re-
significantly increased. Then, a FTC control law in the feed- spectively; iα and iβ are α- and β- axis stator current,
back path is designed. The characteristics of the proposed respectively; ψα and ψβ are α- and β- axis stator flux linkage,
non-smooth composite control method can be outlined below. respectively.
• Simple structure and easy to realize
B. DIRECT TORQUE CONTROL
• Finite-time convergence of output tracking error
To unite the DTC scheme, the PMSMs torque is analysis
• Strong ability of disturbance rejection
from the flux linkage and torque angle. The equation of
For the performance evaluation of presented approach, electromagnetic torque Te of PMSM can be written as
simulation and experimental implementation of PMSM are
3 np
carried out in this paper. It is shown that the effectiveness and Te = |ψs | (2ψr Lq sin δ−|ψs | (Lq −Ld ) sin 2δ) (6)
superiorities of the proposed non-smooth composite control 4 Ld Lq
method are well demonstrated. where δ is the torque angle, which is the electrical angle be-
The rest of this paper is organized as follows. Section II tween stator and rotor flux linkages. As for surface mounted
introduces the system model of PMSM and the principle of PMSM, the Ld = Lq = Ls . Thus, the electromagnet torque
DTC. A non-smooth composite control algorithm and the can be simplified as
stability analysis are presented in section III. In Section IV,
fine-grained simulation and experimental studies have been 3 np
Te = |ψs | ψr sin δ. (7)
done to confirm the validity of the theoretical results. Finally, 2 Ls
a conclusion and reference list end the paper. According to (7), for the purpose of taking full advantage
of the motor quota, if the amplitude of rotor flux linkage
II. PRELIMINARIES ψr remain unchanged, when motor runs, the magnitude of

2 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Author et al.: 11

FIGURE 1. Block diagram of direct torque control for PMSM.

electromagnetic torque Te can be changed by the power angle Remark 1. In most of the existing results [13], only the
δ, which changed with the amplitude of stator flux linkage ψs ideal model is considered, many practical characteristics, like
and the rotating speed ω. This is the essential idea of DTC. friction, sensors errors and inertia variation, are often neglect-
The output of the system is the feedback speed and the ed. Therefore, based on such an ideal model, how to eliminate
PMSM dynamic equation can be further written as: the effects of parametric uncertainties and disturbances is a
major concern. However, if those non-ideal characteristics
1 B 1


 ω̇ = Te − ω − TL are not included in the model of PMSM, it is difficult to


 J J J achieve the desired performances of speed regulation for
1 B 1 1

= Te∗ − ω − TL − (Te∗ − Te ) (8) PMSM.

 J J J J
 = 1 T ∗ + d(t),



III. CONTROLLER DESIGN AND STABILITY ANALYSIS
J e
A. FTDO-BASED CONTROLLER FOR DTC
where Te∗ is the input of reference torque, d(t) = − B Jω − Assumption 1. d(t) in Eq.(8) is first-order differentiable and
1 1 ∗
T
J L − J (T e − Te ) is the lumped disturbances of PMSM ˙ has Lipschitz constancy vector l ∈ Rn .
d(t)
system, which includes the friction, load disturbances, mod- Remark 2. In practice, lots of disturbances satisfy As-
eling error and tracking error. The principle block diagram is sumption1, such as constant disturbances, ramp disturbances,
shown in Fig.1. and harmonic disturbances, etc.
Under Assumption 1, to estimate the disturbance and its
C. DISTURBANCES ANALYSIS OF PMSM derivative, finite-time disturbance observers for system (8)
In practical PMSM servo system, disturbances/uncertainties can be designed as follows:
can be briefly and divided into three categories: paramet- 
ric uncertainties, unmodeled dynamics and external distur-  ṙ1 = ξ + u

bances. ξ = r2 − λ1 l0.5 |r1 − ω|0.5 sign(r1 − ω) (9)
(1) Parametric Uncertainties: Parameters in PMSM sys- 

ṙ2 = −λ2 lsign(r2 − ξ)
tem can be classified as electrical parameters, e.g. stator
resistance and inductance, and mechanical parameters, like T∗
where u = Je , l > 0, λ1 > 0, λ2 > 0. Here, according to the
inertia, friction and rotor flux linkage. analysis in [28], r1 is the estimate of ω and r2 is the estimate
(2) Unmodeled Dynamics: The unmodeled dynamic of of −d(t).
PMSM often includes cogging torque, misalignment of shaft,
broken shaft, measurement error effects, slot harmonics, B. DESIGN OF NON-SMOOTH COMPOSITE
dead-time effects and the unique motor structure, etc. CONTROLLER
(3) External Disturbances: Load torque can be seen as
Define the speed error as:
the major disturbance for PMSM. The position/speed will
inevitably be impacted by load torque variation. In addition, e = ω ∗ − ω. (10)
friction torque is another non-negligible effect in PMSM
system. where ω ∗ is the speed reference. The derivative of (10) is

VOLUME 4, 2016 3

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Author et al.: 11

FIGURE 2. Block diagram of the PMSM control system based on proposed scheme.

Lemma 1. [19] (Finite-time Lyapunov stability) Suppose

1 ∗
ė = ω̇ ∗ −
T − d(t). (11) there exist a C 1 positive definite function V : U → R
J e defined on an open neighborhood U of the origin such that
Combining the disturbance feed-forward compensation V̇ + cV α ≤ 0 on for some c > 0 and x ∈ (0, 1). Then
and the finite-time feedback control technique together, the the origin of system (13) is a finite-time stable equilibrium
reference of electromagnetic torque can be given as 1
and the settling time T satisfies T (x0 ) ≤ c(1−α) V (x0 )1+α
n
Te∗ = J ω̇ ∗ + Kp J(ω ∗ − ω)α − Jr2 . (12) for the initial state x0 . if U = R and V (x) is radically
unbounded, the origin of system (13) is global finite-time
where Kp > 0, 0 < α = q1 /q2 < 1 (q1 and q2 stable.
are positive odd integers). The control diagram of proposed Theorem 1. Assume the desired speed ω ∗ is twice differ-
control method can be found in Fig. 2. entiable. For system (8), if Assumption 1 holds and controller
is (12), the PMSM speed ω will track the desired speed ω ∗ in
C. STABILITY ANALYSIS a finite time, where Kp > 0, 0 < α = q1 /q2 < 1 (q1 and q2
Definition 1. [31] Finite-time stability is studied for a class are positive odd integers).
of autonomous systems, and finite-time stability is exhibiting Proof. The stability analysis for closed loop system (10)
a better rejection of disturbances and uncertainties. Besides, and (12) is shown in the following which could be divided
based on Lyapunov theory, some useful results for finite-time into two steps. In step 1, system states of (10) and (12) will
stability are given as follows. not escape to infinity for t < t1 . In step 2, system (10) and
Considering the following nonlinear autonomous system: (12) is finite time stable when t > t1 .
ẋ = f (x), x ⊆ Rn , f (0) = 0 (13) Step 1 Defining the observer error as ed = d(t) − d(t). ˆ
n
Choosing the Lyapunov Function as
where f : U0 → R is continuous on an open neighbor-
hood U of the origin and f (0) = 0. Then zero solution 1 2
V = e , (14)
of (13) is said to be finite-time convergent if there is an 2
open neighborhood U ⊆ U0 of the origin and a function and the derivative of V can be written as
T : U/{0} → (0, ∞) called the settling time function,
such that ∀x0 ∈ U , the solution x(t, x0 ) = 0. If the zero V̇ = eė
solution of (13) is finite-time convergent, the set of point x0
= e[−Kp eα + ed ] (15)
such that x(t, x0 ) → 0 is called the domain of attraction of
α+1
the solution. The zero solution of (13) is said to be a finite- = −Kp e + ed e.
time stable equilibrium if it is Lyapunov stable and finite-time
Since the d(t) is estimated by d(t) ˆ in finite time t1 ,
convergent. Moreover the zero solution of (13) is said to be
ˆ
respectively, then |d(t) − d(t)| is bounded for t < t1 . Thus,
a globally finite-time stable equilibrium if it is a finite-time
ˆ
there exist constants ε, such that |d(t) − d(t)| ≤ ε. The
stable equilibrium with U = U0 = Rn .
Remark 3. [20] It is clear that from the definition 1, local following inequality can be established
finite-time stability and global asymptotic stability mean
global finite-time stability. V̇ ≤ −Kp eα+1 + εα−1 εα e. (16)

4 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Author et al.: 11

(a) (a)

(b) (b)

(c) (c)
FIGURE 3. Performance comparisons of speed response curves (simulation). FIGURE 4. Performance comparisons of torque response curves (simulation).

According to Young’s inequality [32]: By solving inequality (17), it can be obtained that
α α+1 1 1 3
V̇ ≤ −Kp eα+1 + εα−1 ( ε + eα+1 ) V̇ ≤ −(Kp − C)( + e2 ) + E
α+1 α+1 (17) 2 2
≤ −(Kp − C)eα+1 + E, 1 (19)
≤ −(Kp − C)( + 3V ) + E
α−1
2
where kp > C (C = εα+1 ), E = α+1
α
ε2α , and eα+1 can be = BV + G
written as
eα+1 ≤ |e| + e2 where B = −3(Kp − C) and G = E − 12 (Kp − C). Hence,
1 system state e is bounded when t < t1 .
≤ (1 + e2 ) + e2 (18) Step 2 It will be that closed loop system (10) and (12) is
2
1 3 2 finite time stable when t > t1 . Before proceeding, denote
= + e .
2 2 α+1
eα+1 = (2V ) 2 (20)

VOLUME 4, 2016 5

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Author et al.: 11

and
V̇ ≤ −Kp eα+1
α+1
≤ −Kp (2V ) 2 (21)
β
≤ −bV
β
where β = α+1 2 < 1 and b = 2 Kp . So the (21) can be
further written as

V̇ + bV β ≤ 0, (22)
(a)
∃V makes that
V + αV c ≤ 0. (23)
It can be concluded from Lemma 1 that the closed-loop
system (10) and (12) is finite-time stable, i.e. the feedback
speed ω can track the desired speed ω ∗ in a finite time. This
completes the proof.

TABLE 1. Specification of PMSM

Parameters Meaning Value Unit

Np Number of poles-pairs 4 -
R Stator resistance 9.7 Ω
L Stator inductance 26 kg.m2
ψf Magnetic flux linkage 0.084 Wb
J Moment of inertial 1.35 × 10−4 kg.m2
Bv Friction coefficient 7.4 × 10−5 N.m.s/rad (b)
Ω =Ohm, kg=kilogram, m=meter, N =N ewton, rad=radius, W b =
weber. FIGURE 5. Experimental system. (a) Configuration. (b) Setup.

TABLE 2. Controlling Parameters of Simulation

PI Parameter P+ESO Parameter FTC+FTDO Parameter

Kp 50 Kp 50 Kp 50
Ki 10 p 200 α 0.2
λ1 60
λ2 45

IV. SIMULATION AND EXPERIMENTAL RESULTS

A. SIMULATION RESULTS
In order to verify the effectiveness of the proposed non-
smooth composite method, the simulation test based on
Matlab/Simulink has been performed. The specific parame-
(a)
ter values of PMSM in simulation and experiment are the
same which can been found in Table I, and the controller
parameters, in simulation, are selected as Table II. To have
a comparison, two kinds of typical control algorithms are
employed to make valid comparisons. One is PI control
algorithm, and other one is P+ESO algorithm.
The response curves of speed and torque of PMSM can be
found in Fig.3 and Fig.4, respectively. Fig.3 (a) is the entire
speed response curve, Fig.3(b) is the speed response of start-
up phase and Fig.3 (c) is the speed response with sudden
load. From those figures, in contrast to PI and P+ESO, the
proposed non-smooth composite control has a lower over-
shoot and a shorter recovery time. When a load variation
(TL = 3N.m) is added suddenly, the proposed scheme owns (a)
a less speed drop and shorter adjustment time. Fig.4 (a),(b) FIGURE 6. Performance comparisons at start-up (experiment).
and (c) are corresponding torque response curves.

6 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
Author et al.: 11
(a)
(b)
FIGURE 7. Performance comparisons with sudden disturbance (experiment).
Remark 1: In practical process of adjusting-parameter, a
larger proportional gain of PI is beneficial to convergence
rate and precision, but meanwhile, it will cause larger control
energy consumption and overshoot. Simular to PI, in P+ESO
approach, bigger Kp will lead to higher tracking precision
and faster convergence rate, but larger control energy; and
vice versa. The amplitude of pole p decides the performance
of anti-disturbance, but over large p will lead to the problem
of noise amplification. Notably, in proposed non-smooth
composite control scheme, a larger Kp will obviously result
in a lower steady error and faster convergence speed. In paral-
lel, the parameter α also plays a key role in the gain parameter
settlements, a larger α may cause an apparent deterioration
of the system robustness against the measurement noises.
Hence, in the practices, a proper value of α can be selected
via a trial and error method.
TABLE 3. Controlling Parameters of Experiment
PI Parameter P+ESO Parameter FTC+FTDO
Kp 12 Kp 12 Kp
Ki 5 p 50 α 0.2
λ1
λ2
VOLUME 4, 2016
2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
10.1109/ACCESS.2019.2905103, IEEE Access
B. EXPERIMENTAL RESULTS
To further verify validity of the proposed non-smooth com-
posite controller, the experimental platform for speed regula-
tion of PMSM system are established. The experimental test
setup configuration is illustrated in Fig.5. The control part
of experimental setup adopts the dSPACE DS1104 processor
board, which own the capacity of floating-point computation
by using a 603 Power PC processor and the clock rate up
to 250MHz and offers four 12-bit ADC units and a four-
channel 16-bit (multiplexed) ADC. The power inverter circuit
is comprising of tree parts, 1) single-phase of diode bridge
rectifier, 2) IGBT inverter and 3) large capacitor filter. The
phase current and bus voltage signals can be measured by
Hall sensors, and the position and speed of PMSM can
be obtained by a 17-bit photoelectric shaft encoder. The
controlling parameters of PMSM, in experiment, are selected
as Table III.
In experimental, the response curves of speed and torque
of PMSM at 1000rpm are demonstrated in Fig.6. Similar
to simulation results, in contrast to PI (41ms) and P+ESO
(22ms), the proposed non-smooth composite controller owns
a shorter settling time (12ms) with a small overshoot (1.3%).
The performance comparison of anti-disturbance are shown
in Fig.7. When a load torque (TL = 2N.m) is added sudden-
ly at 1000 rpm by magnetic powder braker, the non-smooth
composite controller has a lower speed drop (20rpm) than
PID control (76rpm) and P+ESO control (42rpm), and gives
a shorter recovery time (20ms) than PID control (140ms)
and P+ESO control (30ms). The experimental results show
that, comparing with typical control methods, such as PID
and P+ESO control, the proposed non-smooth composite
control scheme has significantly enhanced the performance
for disturbance rejection.
To make the comparison more clearer, the detailed perfor-
mance indexes of set-point tracking case are also included in
Table IV.
TABLE 4. Dynamic and steady-state tracking performances
Index PI P+ESO FTC+FTDO
Settling time (ms) 41 22 12
Overshoot (%) 5.2 3.8 1.3
Ripple range (rpm) 17 15 9
V. CONCLUSION
Aiming at performance degradation caused by disturbances
in PMSM system, in this paper, a non-smooth composite
control method has been applied to DTC of PMSM for
improving the dynamic response and capacity of resisting
Parameter disturbance. First, a FTDO are adopted to estimate the
12 lumped disturbances, which include the parameter pertur-
20
bation, friction and load torque, etc. The observed value is
10 employed as a feedforward compensation for lumped distur-
bances of PMSM’s system. The results of simulation and
experiment, in contrast to P+ESO control and PI control,
have demonstrated that the proposed non-smooth compos-
7
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2905103, IEEE Access

Author et al.: 11

ite control scheme has vastly improved the performance of [19] A. Levant, “Higher-order sliding modes, differentiation and output-
PMSM in dynamic response and robustness against distur- feedback control,” International Journal of Control, vol. 76, no. 9-10,
pp. 924–941, 2003.
bances/uncertainties. [20] S. Li, H. Liu, and S. Ding, “A speed control for a pmsm using finite-
Future works will further focus on the design of finite time feedback control and disturbance compensation,” Transactions of the
Institute of Measurement & Control, vol. 32, no. 2, pp. 170–187, 2010.
control in other loops, such as iq -loop and id -loop. Moreover, [21] X. Huang, W. Lin, and B. Yang, “Global finite-time stabilization of a class
taking the disturbance rejection and efficiency optimization of uncertain nonlinear systems,” Automatica, vol. 41, no. 5, pp. 881–888,
into account at the same time is another interesting topic. 2005.
[22] C. Zhang, Y. Yan, A. Narayan, and H. Yu, “Practically oriented finite-
. time control design and implementation: Application to a series elastic
actuator,” IEEE Transactions on Industrial Electronics, vol. 65, no. 5,
pp. 4166–4176, 2018.
REFERENCES [23] Q. Lan, S. Li, S. Khoo, and P. Shi, “Global finite-time stabilisation for
[1] P. Krause, O. Wasynczuk, and S. Sudhoff, “Analysis of electric machinery a class of stochastic nonlinear systems by output feedback,” International
and drive systems,” vol. 21, no. 9, pp. 221–289, 2013. Journal of Control, vol. 88, no. 3, pp. 494–506, 2015.
[2] A. R. I. Mohamed, “Design and implementation of a robust current- [24] S. Bhat and D. S. Bernstein, “Finite-time stability of homogeneous system-
control scheme for a pmsm vector drive with a simple adaptive disturbance s,” in American Control Conference, 1997. Proceedings of the, pp. 2513–
observer,” IEEE Transactions on Industrial Electronics, vol. 54, no. 4, 2514 vol.4, 1997.
pp. 1981–1988, 2007. [25] S. P. Bhat and D. S. Bernstein, “Continuous finite-time stabilization of
[3] I. Shnaps and E. Rimon, “Online coverage by a tethered autonomous the translational and rotational double integrators,” IEEE Transactions on
mobile robot in planar unknown environments,” IEEE Transactions on Automatic Control, vol. 43, no. 5, pp. 678–682, 2002.
Robotics, vol. 30, no. 4, pp. 966–974, 2014. [26] S. P. Bhat, “Finite-time stability of continuous autonomous systems,” Siam
[4] S. Li and Z. Liu, “Adaptive speed control for permanent-magnet syn- J.contr.opti, vol. 38, no. 3, pp. 751–766, 2006.
chronous motor system with variations of load inertia,” IEEE Transactions [27] S. Yu, X. Yu, B. Shirinzadeh, and Z. Man, “Continuous finite-time control
on Industrial Electronics, vol. 56, no. 8, pp. 3050–3059, 2009. for robotic manipulators with terminal sliding mode âŸE,” ˛ in Informa-
[5] R. Errouissi, M. Ouhrouche, W. H. Chen, and A. M. Trzynadlowski, “Ro- tion Fusion, 2003. Proceedings of the Sixth International Conference of,
bust nonlinear predictive controller for permanent-magnet synchronous pp. 1433–1440, 2005.
motors with an optimized cost function,” IEEE Transactions on Industrial [28] J. Wang, C. Zhang, S. Li, J. Yang, and Q. Li, “Finite-time output feedback
Electronics, vol. 59, no. 7, pp. 2849–2858, 2012. control for pwm-based dc-dc buck power converters of current sensorless
[6] S. Chai, L. Wang, and E. Rogers, “A cascade mpc control structure for a mode,” IEEE Transactions on Control Systems Technology, vol. PP, no. 99,
pmsm with speed ripple minimization,” IEEE Transactions on Industrial pp. 1–13, 2016.
Electronics, vol. 60, no. 8, pp. 2978–2987, 2013. [29] Z. Sun, S. Li, and X. Zhang, “Direct torque control of induction machine
using finite-time control and disturbance compensation,” in Industrial
[7] W. C. Gan and L. Qiu, “Torque and velocity ripple elimination of ac per-
Electronics Society, IECON 2014 - Conference of the IEEE, pp. 528–534,
manent magnet motor control systems using the internal model principle,”
2014.
Mechatronics IEEE/ASME Transactions on, vol. 9, no. 2, pp. 436–447,
[30] Z. Sun, a. Y. Guo, Tianliang, X. Wang, and S. Li, “A composite current-
2004.
constrained control for permanent magnet synchronous motor with time-
[8] I. Takahashi and T. Noguchi, “A new quick-response and high-efficiency
varying disturbance,” Advances in Mechanical Engineering, vol. 9, no. 9,
control strategy of an induction motor,” IEEE Transactions on Industry
pp. 1–13, 2017.
Applications, vol. 22, no. 5, pp. 820 – 827, 1986.
[31] Y. Shen, Y. Huang, and J. Gu, “Global finite-time observers for lipschitz
[9] M. Depenbrock, “Direct self-control (dsc) of inverter-fed induction ma- nonlinear systems,” IEEE Transactions on Automatic Control, vol. 56,
chine,” IEEE Transactions on Power Electronics, vol. 3, no. 4, pp. 420– no. 2, pp. 418–424, 2011.
429, 2002. [32] H. Du, S. Li, and C. Qian, “Finite-time attitude tracking control of space-
[10] I. C. Baik, K. H. Kim, and M. J. Youn, “Robust nonlinear speed control of craft with application to attitude synchronization,” IEEE Transactions on
pm synchronous motor using boundary layer integral sliding mode control Automatic Control, vol. 56, no. 11, pp. 2711–2717, 2011.
technique,” Control Systems Technology IEEE Transactions on, vol. 8,
no. 1, pp. 47–54, 2000.
[11] S. Bolognani, S. Bolognani, L. Peretti, and M. Zigliotto, “Design and
implementation of model predictive control for electrical motor drives,”
IEEE Transactions on Industrial Electronics, vol. 56, no. 6, pp. 1925–1936,
2009.
[12] H. Chaoui and P. Sicard, “Adaptive fuzzy logic control of permanent
magnet synchronous machines with nonlinear friction,” IEEE Transactions
on Industrial Electronics, vol. 59, no. 2, pp. 1123–1133, 2011.
[13] J. Yang, W. H. Chen, and S. Li, “Non-linear disturbance observer-based
robust control for systems with mismatched disturbances/uncertainties,”
Iet Control Theory & Applications, vol. 5, no. 18, pp. 2053–2062, 2011.
[14] J. Davila, L. Fridman, and A. Levant, “Second-order sliding-mode ob-
server for mechanical systems,” IEEE Transactions on Automatic Control,
vol. 50, no. 11, pp. 1785–1789, 2005.
[15] J. H. She, M. Fang, Y. Ohyama, H. Hashimoto, and M. Wu, “Im-
proving disturbance-rejection performance based on an equivalent-input-
disturbance approach,” IEEE Transactions on Industrial Electronics,
vol. 55, no. 1, pp. 380–389, 2008.
[16] S. J. Kwon and K. C. Wan, “A discrete-time design and analysis of
perturbation observer for motion control applications,” IEEE Transactions
on Control Systems Technology, vol. 11, no. 3, pp. 399–407, 2004.
[17] T. G. Park and D. Kim, “Design of unknown input observers for linear
systems with unmatched unknown inputs,” Transactions of the Institute of
Measurement & Control, vol. 36, no. 3, pp. 399–410, 2014.
[18] X. Li and S. Li, “Speed control for a pmsm servo system using model
reference adaptive control and an extended state observer,” Journal of
Power Electronics, vol. 14, no. 3, pp. 549–563, 2014.

8 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.