Received June 19, 2018, accepted July 19, 2018, date of publication August 7, 2018, date of current version

August 28, 2018.

Digital Object Identifier 10.1109/ACCESS.2018.2862637

Port-Controlled Hamiltonian and Sliding

Mode Control of Gantry Robot Based
on Induction Motor Drives
BINGKUN ZHAO, HAISHENG YU , JINPENG YU, XUDONG LIU , AND HERONG WU
College of Automation and Electrical Engineering, Qingdao University, Qingdao 266071, China
Corresponding author: Haisheng Yu (yu.hs@163.com)
The work was supported in part by the Natural Science Foundation of China under Grant 61573203, Grant 61573204, and Grant 61703222,
and in part by the Shandong Province Outstanding Youth Fund under Grant ZR2015JL022.

ABSTRACT A port-controlled Hamiltonian (PCH) control approach is presented to solve the position
tracking problem of gantry robot based on induction motor (IM) drives. First, a robot model is established.
Second, a PCH controller is designed to realize accurate position tracking of a gantry robot. For IM drives,
it is convenient to choose a direct torque control strategy based on the sliding mode control, which overcomes
the higher ripples of torque and flux. Third, a voltage reconstruction technique is introduced to calculate the
stator voltage of the IM, which replaces the stator voltage measurement of the IM. Finally, the load torque
observer is developed to estimate an unknown load torque. The asymptotic stability of the robot system is
proved by the Lyapunov stability theory. Simulation results indicate that the system has excellent position
tracking performances and load disturbance attenuation ability.

INDEX TERMS Gantry robot, port-controlled Hamiltonian, induction motor, sliding mode control, load
torque observer.

I. INTRODUCTION control has attracted lots of attentions and is widely applied

The gantry robot is widely used in industry applications. The to the nonlinear systems [12], [13]. The dynamics model of
techniques of robots control are well treated in the literature, gantry robot can also be viewed as a PCH system and then
such as back-stepping control [1], adaptive Neuro-sliding the PCH controller can be obtained by damping injection and
mode control [2]. The back-stepping controller generates the energy-shaping [14], [15]. So the PCH controller is designed
desired control force on the mechanical subsystem. The adap- to realize accurate position tracking of gantry robot.
tive Neuro-sliding mode control method has a good steady For the drive systems of gantry robot, the induction
tracking accuracy without chattering in the control signal. motor (IM) is broadly used in servo systems since it has the
The position tracking issues of the robots are increasingly advantages of low price and ruggedness. Therefore, the IM is
being concerned [3], [4]. Some simple methods are emploied suitable for the application in the robot. Field oriented con-
to position control of robots, such as gravity compensation [5] trol (FOC) and direct torque control (DTC), appear to be very
and proportion differentiation (PD) control [6]. While the convenient for good dynamic response. The DTC strategy
fixed parameters of PD control can not adapt the variability of based on hysteresis controllers and lookup switching table
system. Sometimes robot manipulators are difficult to estab- has more advantages such as fast response and less depen-
lish an accurate mathematical model due to robot’s uncer- dence to the parameters of IM. However, the main problems
tainties and nonlinear characteristics [7]. For this reason, of DTC are the high torque and flux ripples, as well as the
a large number of intelligent control strategies are applied to variable switch frequency, which increase control difficulty
the gantry robot drive systems. For instance, adaptive neural in low speed regions [16], [17]. Nowadays, there are different
network [8], [9], adaptive fuzzy neural network [10], and approaches being used to cope with these disadvantages.
RBF neural network [11]. But the intelligent control relies on In [18], a multilevel converter is provided to reduce the rip-
numerous programming languages that are complex and cum- ples. Nevertheless, the number of power switches is increased
bersome. Note that the Port-Controlled Hamiltonian (PCH) which causes the lower efficiency of the control system.

2169-3536 2018 IEEE. Translations and content mining are permitted for academic research only.
43840 Personal use is also permitted, but republication/redistribution requires IEEE permission. VOLUME 6, 2018
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

 T
The model predictive control (MPC) also has the ability to where q = qx qy qz , di are the angular position and gear
determine a suitable voltage vector which can reduce the rotation radius of load shaft, respectively, i = x, y, z.
ripples [19]. Nevertheless, the high complexity of the control The drive structure of i-axis is shown in Fig.2.
law is the main drawback of MPC. For IM drives, it is
convenient to choose direct torque control (DTC) strategy
based on the sliding mode control (SMC), which overcomes
the higher ripples of torque and flux, as well as the variable
switch frequency. The sliding mode control (SMC) has good
robustness and fast response, thus it can be designed to con-
trol the torque and flux of IM. According to [20] and [21],
the torque and flux sliding mode controllers are developed
to reduce the ripples. Then, a sliding mode direct torque
control (SM-DTC) strategy is presented to the control of
the IM. The robot control combined with motor control is
closer to the actual industry application. And the torque is the FIGURE 2. The drive structure of i -axis.
connection between the robot and induction motors. At the
same time, the desired torque obtained from PCH controller where i = x, y, z, Jai , Ji are the inertia moments of motor
can be the input of induction motors directly instead of the and load, respectively. τi , Ti are the torque of motor and load,
torque current conversion in vector control. The contributions respectively. θi , qi are angular positions of motor and load,
of the article are: respectively. ri , rmi are the friction coefficients of motor and
(1) The PCH control scheme is developed to enhance the load, respectively. ki is the reduction ratio between motor and
position tracking performance of gantry robot. load. Thus, the dynamics equation of i-axis between motor
(2) Compared with [22] and [23], a novel DTC strategy and load is calculated as follows
based on the sliding mode control is investigated which
(Jai + Ji ki2 )θ̈i + (ri + rmi ki2 )θ̇i = τi . (2)
decreases the torque and flux ripples. At the same time,
a voltage reconstruction technique is proposed to calculate the The robot dynamics model can be described as
stator voltage, which replaces the stator voltage measurement
of the IM. M (θ)θ̈ + C(θ, θ̇ )θ̇ + Rf θ̇ + G = τ (3)
(3) The robot system has good load disturbance attenuation where θ, θ̇ , θ̈ are 3 × 1 angular position, velocity and accel-
performance by introducing the load torque observer. eration vectors, respectively, M (θ ) is a 3 × 3 positive definite
(4) The robot control combined with motor control is and symmetric mixed inertia matrix, G is a 3 × 1 gravity
closer to the actual industry application. The whole system
is asymptotically stable based on Lyapunov stability theory.
torque T τ represents a 3 × 1 input torque and τ =
vector,
τx τy τz , Rf is friction coefficient, C(θ, θ̇ ) is a 3 × 1


matrix of centripetal and Coriolis forces.

II. MODEL OF GANTRY ROBOT SYSTEM n o
A. KINEMATICS AND DYNAMICS MODELS M (θ) = diag Jax + Jx kx2 , Jay + Jy ky2 , Jaz + Jz kz2 ,
OF GANTRY ROBOT T
Rf = rx + rmx kx2 ry + rmy ky2 rz + rmz kz2 ,

The gantry robot has three axes, where x, y axis are in the
horizontal plane, and z axis is vertical to the horizontal level. T
G = 0 0 mz gkz dz /(2π) , C(θ, θ̇) = 0.

Three axes are perpendicular to eath other. The IM is chosen
as the drive motor. The robot model is shown in Fig.1 and B. THE MATHEMATICAL MODEL OF THE IM
base coordinate system is set at point ’o’.
The model of the IM can be expressed in αβ frame. It is
expressed as [24]
Rri λsαi

Rsi Lri + Rri Lsi
 i̇sαi = − isαi − ωri isβi + +
σ σ

L L Lsi Lri

si ri

ω λ


 ri sβi
 u sαi
 +
 σ Lri σ Lsi


Rri λsβi

 Rsi Lri + Rri Lsi
isβi + ωri isαi +

 i̇sβi = − −

σL L σ Lsi Lri


FIGURE 1. The model of gantry robot. ωri usβi si ri (4)
+

 σ L ri σ Lsi
λ̇sαi = usαi − Rsi isαi



The kinematics model can be presented as 

λ̇sβi = usβi − Rsi isβi


dx /2π τi τLi
     
x 0 0 qx


ω̇i = −

dy /2π

y= 0 0   qy  (1)


 Ji Ji
dz /2π θ̇i = ωi

z 0 0 qz


VOLUME 6, 2018 43841

 B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

where isαi , isβi are stator current components, respectively, to decrease the ripples of torque and flux. On the other
λsαi , λsβi are stator flux components, respectively, Rsi , Rri are hand, a PCH controller is applied as position regudation.
stator and rotor resistances, respectively, Lsi , Lri are stator and What’s more, a load torque obsever is added to τi∗ and then
L2 τi0∗ = τi∗ + τ̂Li .
rotor inductances, respectively, σ = 1 − LsimiLri is Blondel’s
coefficient, Lmi is the mutual statorrotor inductance, Ji is
inertia moment, θi , ωi are the angular position and velocity, C. PCH CONTROLLER DESIGN
respectively. The i is a subscript and i = x, y, z. To improve the position tracking performance, the PCH con-
The electromagnetic torque can be presented as trol strategy is introduced. Then the energy shaping and
damping injection methods are proposed to obtain the torque
τi = np (λsαi isβi − λsβi isαi ) (5) for the IM.
qi = ki θi
T
Define p = M (θ)θ̇ as system momenta, η = θ p as the

(6)
where np is the number of pole pairs. state vector, then θ̇ = M −1 (θ )p and V (η) = mz gkz dz θz /(2π ).
The Hamiltonian system of the robot is expressed as
III. CONTROLLER DESIGN 1 T −1
A. GANTRY ROBOT CONTROL SYSTEM H (η) = η M (θ )η + V (η). (7)
2
According to inverse kinematics and dynamics model of
robot, the whole servo control system of robot is shown The Hamiltonian system of gantry robot is shown in the
in Fig.3. λ∗sx , λ∗sy and λ∗sz are the constant fluxes refer- following [25]
ence (0.8Wb). After the desired position (x ∗ , y∗ , z∗ ) is given, ∂H (η)
the reference angle position (q∗x , q∗y , q∗z ) of robot is obtained η̇ = (J − R) + g(η)τ (8)
∂η
by inverse kinematics. Then the reference torque τx∗ , τy∗ and
where R represents the dissipation, the interconnection struc-
τz∗ can be calculated by PCH controller which is designed
ture is captured in matrix g(η) and the skew-symmetricmatrix
based on position errors. Finally, the SMC strategy based on
J = −J T .
torque and flux errors is developed to motor control subsys-    
tems. 0 I 0 0 T
T
, R= , g(η) = 0 I .

J = −J =
−I 0 0 Rf
Let the desired Hamiltonian function of robot system as
follows
1 1
Hd (η) = (η − η∗ )T Md−1 (θ )(η − η∗ ) + ρ(η − η∗ )2 (9)
2 2
where, M (θ ) = Md (θ) = MdT (θ ) > 0, and Vd (η − η∗ ) =
2 ρ (η − η ) represents the potential energy function, ρ(>
1 ∗ 2

0) is a parameter. Thus Vd (η − η∗ ) has

T an minimum at the
desired equilibrium point η∗ = θ ∗ 0
θ ∗ = arg min Vd (η − η∗ ). (10)
The desired closed-loop Hamiltonian system can be
described as
FIGURE 3. Gantry robot control system schematic diagram. ∂Hd (η)
η̇ = (Jd − Rd ) (11)
∂η
where, Kv is the designed parameter, the desired interconnec-
tion matrix and damping matrix are
   
0 I 0 0
Jd = −JdT = , Rd = RTd = ≥ 0.
−I 0 0 Kv
The PCH position controller is obtained from Eq.(8) and
Eq.(11), that is
∂H (η)
(J − R) + g(η)τ
FIGURE 4. The DTC of IM based on SMC. ∂η
∂Hd (η)
= (Jd − Rd )
B. THE I-AXIS CONTROL SYSTEM ∂η
∂H ∂H (η) ∂Hd (η) ∂Hd (η)
 
(η)
The i-axis control system of the robot is expressed in Fig.4. τ =
∗
+ Rf − − Kv (12)
On the one hand, the SMC is designed as inner loop controller ∂θ ∂p ∂θ ∂p

43842 VOLUME 6, 2018

B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

therefore, τx∗ , τy∗ and τz∗ can be calculated as follows The positive Lyapunov function can be indicated in the
 following
Kv px

 τx∗ = (rx + rm ki2 )θ̇x − ρ(θx∗ − θx ) − 2 V2 = wT Pw (17)
(Jax + Jx )ki




Kv py


τy∗ = (ry + rm ki2 )θ̇y − ρ(θy∗ − θy ) − where P is a positive definite matrix. Bythe Lyapunov equa-

 (Jay + Jy )ki2 tion AT P + PA = −Q (Q = diag 1 1 1 ), P can be
mz gkdz Kv pz

 calculated as follows
 τz = +(rz +rm ki2 )θ̇z −ρ(θz∗ −θz ) −


 ∗
2π (Jaz + Jz )ki2 Ji k2i 1
 
0
qi = ki θi , qi = ki θi , i = x, y, z.
∗ ∗  2k3i + 2k1i k2i Ji 2k3i + 2k1i k2i Ji 
 Ji 
P = 0 0
 
Define V1 = Hd (η) as the Lyapunov function, the deriva-  2k3i + 2k1i k2i Ji


tive of V1 along trajectories of Eq.(11) can be written as  1 −k1 
0
follows 2k3i + 2k1i k2i Ji (2k3i + 2k1i k2i Ji )k3i
∂Hd (η) T ∂Hd (η) T ∂Hd (η) we can get k1i > − kk2i3iJi , k2i > 0, k3i < 0 by judging
     
V̇1 = η̇ = − (Jd − Rd )
∂η ∂η ∂η |P|11 > 0, |P|22 > 0, |P|33 > 0 so that P is a positive definite
T
∂Hd (η) ∂Hd (η)
  
matrix. Therefore, we have
=− Rd ≤ 0. (13)
∂η ∂η V̇2 < 0. (18)
∂Hd (η) ∂ 2 Hd (η)
if η = η∗ ,
then = 0 and
∂η ≥ 0. ∂η2 According to the Lyapunov stability theorem, the load
According to LaSalle’s invariance set principle, when torque observer subsystem can be asymptotically stable. Thus
when τLi is unknown, τ̂Li can replace τLi . The characteristic
 T 
η ∈ Rn ∂H∂η Rd ∂H∂η
  
d (η) d (η)
= 0 , the PCH position equation of Eq.(16) is given as
subsystem is asymptotically stable at the desired equilibrium
s3 + k1i s2 + k2i s − k3i /Ji = 0. (19)
point η∗ .
All the poles of the observer are set to be s1 = s2 = s3 =
D. LOAD TORQUE OBSERVER DESIGN sp (< 0) by selection of parameters as
In order to estimate the unknown load torque, the load torque
k1i = −3sp , k2i = 3s2p , k3i = Ji s3p .
observer is developed [26] ,[27].
When the load torque is known and constant, from Eq.(4)
E. VOLTAGE RECONSTRUCTION
we get
The stator voltage reconstruction strategy based on DC
θ̇i = ωτii

 voltage and inverter switch signal is proposed [28], which
ω̇i = − τJLii (14) replaces the stator voltage measurement of the IM. In Fig.5,
 Ji the state of each switch in the inverter is shown by
τ̇Li = 0

s1 , s2 , s3 , s4 , s5 , s6 and udc is measurable. When the upper
However, the load torque is uncertain in actual indus- bridge switch is turned on, s1 =1 (or s2 , s3 ), the lower
try applications. Thus, a load torque observer equation is bridge switch is turned on, s1 =0 (or s2 , s3 ), therefore the
designed as two switches of the same bridge are on alternatively. Then,
 applying SVPWM signal transformation principle, we can
˙
θ̂i = ω i + k1i (θi − θ̂i )
 get six pulses which are used to control the power converter
τi

ω̂i = − τ̂JLii + k2i (θi − θ̂i )
˙ (15) to provide the desired three-phase voltages. The different
 Ji instantaneous values ua , ub , uc can be obtained by selecting
˙
τ̂Li = k3i (θi − θ̂i )

different switch variables s1 , s2 and s3 .
The state average model of the inverter is showed as
where, k1i , k2i , k3i are designed parameters. Define θ̃i = θi −
T  udc
θ̂i , ω̃i = ωi − ω̂i , τ̃Li = τLi − τ̂Li and w = θ̃i ω̃i τ̃Li as

 ua = (2s1 − s2 − s3 )
u3dc


the estimated errors and state error vector, respectively. The 
ub = (2s2 − s1 − s3 ) (20)
error equation of observer has the form 3
u

 dc
 uc = (2s3 − s1 − s2 )

ẇ = Aw (16) 3
The stator voltage in αβ frame is
where
1 1
  

−k1i 1 0
   r 1 − − ua
usα 2 √2 √2   ub .
− J1i  = (21)

A =  −k2i 0 usβ 3
 3 3 
−k3i 0 0 0 − uc
2 2
VOLUME 6, 2018 43843
 B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

According Eq.(26), Eq.(27), Eq.(28) and Eq.(29), the value

of A, B can be calculated below
c1i s1i + ε1i sgn(s1i ) + τ̇i∗
A= − ωri (isαi λsαi + isβi λsβi )
np
Rsi Lri + Rri Lsi ωri 2
− (isαi λsβi − isβi λsαi ) + λ , (30)
σ Lsi Lri σ Lsi si
c2i s2i + ε2i sgn(s2i ) + λ̇2∗
B= si
+ Rsi (isαi λsαi − isβi λsβi ).
2
(31)
FIGURE 5. Inverter and motor block diagram. The sliding mode controller can be calculated by Eq.(30)
and Eq.(31)
σ Lsi λsβi A − (λsαi − σ Lsi isαi )B
The reconstructed stator voltage can be calculated by u∗sαi = , (32)
Eq.(21). The stator flux amplitude and phase angle are cal- σ Lsi (isβi λsβi − isαi λsαi ) − λ2si
culated as follows σ Lsi λsαi A − (σ Lsi isβi − λsβi )B
u∗sβi = . (33)
λsαi = (usαi − Rsi isαi )dt λ2si − σ Lsi (isαi λsαi + isβi λsβi )
 R


 λsβi = (usβi − Rsi isβi )dt
 R
To prove the asymptotic stability of the sliding mode sub-

(22) system, define s = [ s1i s2i ]T as the state vector and let
 λ̂2si = λ2sαi + λ2sβi
V3 = 12 sT s be the Lyapunov function. Therefore the deriva-

 ϕ = arctan(λsβi /λsαi ).


tive of V3 is calculated in the following
The estimated electromagnetic torque is given as 
c1i s1i + ε1i sgn(s1i )

T
V̇3 = s ṡ = −[ s1i s2i ]
τ̂i = np (λsαi isβi − λsβi isαi ). (23) c2i s2i + ε2i sgn(s2i )
= −c1i s21i − ε1i sgn(s1i )s1i − c2i s22i − ε2i sgn(s2i )s2i
F. SMC DESIGN < −c1i s21i − ε1i |s1i | − c2i s22i − ε2i |s2i | < 0 (34)
To simplify the design of the SMC controllers [29], [30],
the errors of the torque and flux are taken as their sliding if and only if s1i = s2i = 0, V̇3 = 0, thus the sliding mode
surfaces, respectively. subsystem is asymptotically stable.

s1i = e1i = τi0∗ − τ̂i (24) G. STABILITY ANALYSIS

s2i = e2i = λ2∗
si − λ̂2si (25) The Lyapunov function of whole system is chosen as V , and
V = V1 + V2 + V3 . According to Eq.(13), Eq.(18), Eq.(34),
the derivatives of Eq.(24) and Eq.(25) are expressed as
and based on the Lyapunov stability theory, then V > 0,
follows
V̇ < 0, thus the whole system is asymptotically stable.
ṡ1i = τ̇i0∗ − npi [λ̇sαi isβi + λsαi i̇sβi − λ̇sβi isαi − λsβi i̇sαi ]
= τ̇i0∗ − npi [A + ωri (isαi λsαi + isβi λsβi ) IV. SIMULATION RESULTS
The simulations are carried out with the following parame-
Rsi Lri + Rri Lsi ωri 2
+ (isαi λsβi − isβi λsαi ) − λ ] ters of induction motor and robot: Rs = 0.96419, Rr =
σ Lsi Lri σ Lsi si 0.93766, Lr = 6.43858mH, Ls = 6.08925mH, np = 2,
(26)
Lm = 5.9mH, Pn = 1.5KW, Un = 220V, and mi = 1kg,
si − λ̇si = λ̇si − 2B + 2Rsi (isαi λsαi − isβi λsβi )
ṡ2i = λ̇2∗ 2 2∗
ri = 0.01, rmi = 0.1, Jai = 0.03kg.m2 , Ji = 0.1kg.m2 ,
(27) di = 0.5cm, ki = 10 and the robot operating space is
(3×3×3)m. The parameters of PCH and SMC controllers are
where Kv = 2, ρ = 2,c1i = 40000, c2i = 3000, ε1i = 1, ε2i = 1.
λsβi λsαi The parameters of PD control are Kp = 40, KD = 3.
A = (isβi − )usαi + ( − isαi )usβi ,
σ Lsi σ Lsi Case 1: The (0,0,0) and (1m, 1.2m, 1.4m) are initial and
B = usαi λsαi + usβi λsβi . desired positions, respectively. In this case, 2N.m load torque
is inserted at t=0s and then 10N.m load torque is added at
The exponential reaching laws of sliding mode controller
t=2s. Fig.6 to Fig.15 give the position tracking performance
are designed as
of robot and the responses of the IM. It can be seen from
ṡ1i = −c1i s1i − ε1i sgn(s1i ) (28) Fig.6 and Fig.7 that the proposed PCH and SMC meth-
ṡ2i = −c2i s2i − ε2i sgn(s2i ) (29) ods have better position tracking performance. Moreover,
Fig.8 and Fig.9 show the speed responses of robot. It is
where c1i , ε1i , c2i , ε2i > 0 . obvious that PCH and SMC can keep better speed responses

43844 VOLUME 6, 2018

B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

FIGURE 9. Speed curves of PCH and SMC control.

FIGURE 6. Position curves of PD and hysteresis control.

FIGURE 10. Electromagnetic torque curves of PD and hysteresis control.

FIGURE 7. Position curves of PCH and SMC control.

FIGURE 11. Electromagnetic torque curves of PCH and SMC control.

FIGURE 8. Speed curves of PD and hysteresis control.

than PD control. From Fig.11, we can know that the proposed control via Fig.12 when the IM moves with a constant veloc-
PCH and SMC controllers have good torque dynamics. The ity. Fig.13 presents that the reconstructed stator voltage can
stator flux has better performance than traditional hysteresis replace the measured stator voltage. At t=2s, different sp

VOLUME 6, 2018 43845

 B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

FIGURE 15. Space motion trajectory.

FIGURE 12. Flux α curves.

FIGURE 13. Voltage reconstruction curves. FIGURE 16. Position curves of PD and hysteresis control.

FIGURE 14. Load torque observation curve. FIGURE 17. Position curves of PCH and SMC control.

values (sp = −100, sp = −60, sp = −40) are added to strategies have smoother trajectories than classical PD and
the load torque observation, Fig.14 illustrates that there is hysteresis control.
a better unknown load torque tracking performance when Case 2: A trapezoidal position signal is applied to the
sp = −100. Finally, Fig.15 show that the PCH and SMC robot. Before t=1s the space position is (1m, 1.2m, 1.4m),

43846 VOLUME 6, 2018

B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

FIGURE 18. Speed curves of PD and hysteresis control. FIGURE 21. Space motion trajectory of PCH and SMC control.

FIGURE 19. Speed curves of PCH and SMC control. FIGURE 22. Position curves of PD and hysteresis control.

FIGURE 20. Space motion trajectory of PD and hysteresis control.

FIGURE 23. Position curves of PCH and SMC control.

then at t=1s steps to (2m, 2.4m, 2.8m), at t=2s back to

(1m, 1.2m, 1.4m). Fig.16 to Fig.19 show that PCH and SMC
π
have better position and speed responses. Fig.20 and Fig.21 Case 3: x ∗ = sin ( 2π ∗ 2π ∗
5 t)rad,y = sin ( 5 t + 6 )rad,z =
2π π
present that PCH and SMC strategies have smoother sin ( 5 t + 3 )rad, which are periodic reference signals.
trajectories. Fig.22 to Fig.25 express the curves of position tracking,

VOLUME 6, 2018 43847

 B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

[2] S. Frikha, M. Djemel, and N. Derbel, ‘‘A new adaptive neuro-sliding mode
control for gantry crane,’’ Int. J. Control, Automat. Syst., vol. 16, no. 2,
pp. 559–565, Apr. 2018, doi: 10.1007/s12555-017-0070-x.
[3] Y. Zhang, P. Yan, and Z. Zhang, ‘‘High precision tracking control of a
servo gantry with dynamic friction compensation,’’ ISA Trans., vol. 62,
pp. 349–356, May 2016, doi: 10.1016/j.isatra.2016.02.006.
[4] W. Yin, L. Sun, M. Wang, and J. Liu, ‘‘Nonlinear state feedback position
control for flexible joint robot with energy shaping,’’ Robot. Auton. Syst.,
vol. 99, no. 1, pp. 121–134, Jan. 2018, doi: 10.1016/j.robot.2017.10.007.
[5] V. Arakelian, ‘‘Gravity compensation in robotics,’’ Adv. Robot., vol. 30,
no. 2, pp. 79–96, 2016.
[6] P. R. Ouyang, V. Pano, J. Tang, and W. H. Yue, ‘‘Position domain non-
linear PD control for contour tracking of robotic manipulator,’’ Robot.
Comput.-Integr. Manuf., vol. 51, pp. 14–24, Jun. 2018, doi: 10.1016/j.rcim.
2017.11.017.
[7] A. Arian, B. Danaei, H. Abdi, and S. Nahavandi, ‘‘Kinematic and dynamic
analysis of the gantry-tau, a 3-DoF translational parallel manipulator,’’
Appl. Math. Model., vol. 51, pp. 217–231, Jun. 2017, doi: 10.1016/
j.apm.2017.06.012.
FIGURE 24. Space motion trajectory of PD and hysteresis control. [8] W. He, Y. Dong, and C. Sun, ‘‘Adaptive neural impedance control of
a robotic manipulator with input saturation,’’ IEEE Trans. Syst., Man,
Cybern. Syst., vol. 46, no. 3, pp. 334–344, Mar. 2016.
[9] W. He, Y. Chen, and Z. Yin, ‘‘Adaptive neural network control of an
uncertain robot with full-state constraints,’’ IEEE Trans. Cybern., vol. 46,
no. 3, pp. 620–629, Mar. 2016.
[10] R.-J. Wai, Y.-C. Huang, Z.-W. Yang, and C.-Y. Shih, ‘‘Adaptive fuzzy-
neural-network velocity sensorless control for robot manipulator posi-
tion tracking,’’ IET Control Theory Appl., vol. 4, pp. 1079–1093,
Jun. 2010.
[11] L. Yu, S. Fei, J. Huang, and Y. Gao, ‘‘Trajectory switching control of
robotic manipulators based on RBF neural networks,’’ Circuits Syst. Signal
Process., vol. 33, no. 4, pp. 1119–1133, Apr. 2014.
[12] H. Yu, J. Yu, J. Liu, and Y. Wang, ‘‘Energy-shaping and L2 gain disturbance
attenuation control of induction motor,’’ Int. J. Innov. Comput. Inf. Control,
vol. 8, no. 7, pp. 5011–5024, Jul. 2012.
[13] H. Yu, J. Yu, J. Liu, and Q. Song, ‘‘Nonlinear control of induction motors
based on state error PCH and energy-shaping principle,’’ Nonlinear Dyn.,
vol. 72, nos. 1–2, pp. 49–59, Apr. 2013.
[14] R. Ortega, F. Gomez-Estern, G. Blankenstein, and M. W. Spong, ‘‘Stabi-
lization of a class of underactuated mechanical systems via interconnection
and damping assignment,’’ IEEE Trans. Autom. Control, vol. 47, no. 8,
FIGURE 25. Space motion trajectory of PCH and SMC control.
pp. 1218–1233, Aug. 2000.
[15] J. Chi, H. Yu, and J. Yu, ‘‘Hybrid tracking control of 2-DOF SCARA robot
space motion trajectory. From Fig.22 to Fig.23, the reference via port-controlled hamiltonian and backstepping,’’ IEEE Access, vol. 6,
pp. 17354–17360, 2018, doi: 10.1109/ACCESS.2018.2820681.
position is tracked by actual position, which shows that the
[16] I. M. Alsofyani and N. R. N. Idris, ‘‘Simple flux regulation for improving
PCH and SMC control strategies have excellent performance. state estimation at very low and zero speed of a speed sensorless direct
torque control of an induction motor,’’ IEEE Trans. Power Electron.,
V. CONCLUSION vol. 31, no. 4, pp. 3027–3035, Apr. 2016.
In view of gantry robot systems, a novel method of PCH and [17] C. Lascu, I. Boldea, and F. Blaabjerg, ‘‘A modified direct torque control for
induction motor sensorless drive,’’ IEEE Trans. Ind. Appl., vol. 36, no. 1,
SMC is developed in this paper. The PCH controller is pro-
pp. 122–130, Jan. 2000.
posed by damping injection and energy-shaping. The desired [18] N. N. Venkataramana, A. Panda, and S. P. Singh, ‘‘A three-level fuzzy-2
equilibrium point is also determined. With the designed DTC of induction motor drive using SVPWM,’’ IEEE Trans. Ind. Electron.,
PCH controller, the gantry robot position control is achieved. vol. 63, no. 3, pp. 1467–1479, Mar. 2016.
[19] M. Habibullah, D. D.-C. Lu, D. Xiao, and M. F. Rahman, ‘‘A simplified
For the drive systems of gantry robot, the DTC strategy finite-state predictive direct torque control for induction motor drive,’’
based on SMC is applied to the IM drive systems, which IEEE Trans. Ind. Electron., vol. 63, no. 6, pp. 3964–3975, Jun. 2016.
reduces the torque and flux ripples. Moveover, the robot has [20] A. Ammar, A. Benakcha, and A. Bourek, ‘‘Closed loop torque SVM-DTC
based on robust super twisting speed controller for induction motor drive
good disturbance attenuation ability by introducing the load with efficiency optimization,’’ Int. J. Hydrogen Energy, vol. 42, no. 28,
torque observer. Simulations show that the proposed control pp. 17940–17952, Jul. 2017.
methods have good position tracking performance and fast [21] C. M. R. Oliveira, M. L. Aguiar, J. R. B. A. Monteiro, W. C. A. Pereira,
G. T. Paula, and T. E. P. Almeida, ‘‘PI control of induction motor using an
speed responses under various conditions. In the future work, integral sliding mode controller with anti-windup,’’ J. Control, Automat.
we will pay attention to the practical application of the pro- Elect. Syst., vol. 8, no. 7, pp. 169–178, Apr. 2016.
posed control strategy. [22] A. Ammar, A. Bourek, and A. Benakcha, ‘‘Sensorless SVM-direct torque
control for induction motor drive using sliding mode observers,’’ J. Con-
trol, Automat. Elect. Syst., vol. 28, no. 2, pp. 189–201, Apr. 2017.
REFERENCES [23] A. Ammar, A. Bourek, and A. Benakcha, ‘‘Nonlinear SVM-DTC for
[1] C. F. Baicu, C. D. Rahn, and D. M. Dawson, ‘‘Backstepping boundary induction motor drive using input-output feedback linearization and
control of flexible-link electrically driven gantry robots,’’ IEEE/ASME high order sliding mode control,’’ ISA Trans., vol. 67, pp. 428–442,
Trans. Mechatronics, vol. 3, no. 1, pp. 60–66, Mar. 1998. Mar. 2017.

43848 VOLUME 6, 2018

B. Zhao et al.: PCH and SMC of Gantry Robot Based on IM Drives

[24] H. Sudheer, S. F. Kodad, and B. Sarvesh, ‘‘Improvements in direct torque JINPENG YU received the B.Sc. degree in
control of induction motor for wide range of speed operation using automation from Qingdao University, Qingdao,
fuzzy logic,’’ J. Elect. Syst. Inf. Technol., to be published, doi: 10.1016/ China, in 2002, the M.Sc. degree in system engi-
j.jesit.2016.12.015. neering from Shandong University, Jinan, China,
[25] R. Ortega, A. J. Van der Schaft, I. Mareels, and B. Maschke, ‘‘Putting in 2006, and the Ph.D. degree from the Institute of
energy back in control,’’ IEEE Control Syst. Mag, vol. 21, no. 2, Complexity Science, Qingdao University, in 2011.
pp. 18–33, Apr. 2001. He is currently a Distinguished Professor with the
[26] L. Yu, J. Huang, and S. Fei, ‘‘Sliding mode switching control of manip-
School of Automation and Electrical Engineering,
ulators based on disturbance observer,’’ Circuits, Syst., Signal Process.,
Qingdao University. His research interests include
vol. 36, no. 6, pp. 2574–2585, Jun. 2017.
[27] A. Mohammadi, M. Tavakoli, H. J. Marquez, and F. Hashemzadeh, ‘‘Non- electrical energy conversion and motor control,
linear disturbance observer design for robotic manipulators,’’ Control Eng. applied nonlinear control, and intelligent systems. He was a recipient of the
Pract., vol. 21, no. 3, pp. 253–267, Mar. 2013. Shandong Province Taishan Scholar Special Project Fund and the Shandong
[28] A. Rahoui, A. Bechouche, H. Seddiki, and D. O. Abdeslam, ‘‘Grid voltages Province Fund for Outstanding Young Scholars.
estimation for three-phase PWM rectifiers control without AC voltage sen-
sors,’’ IEEE Trans. Power Electron., vol. 33, no. 1, pp. 859–875, Jan. 2018.
[29] A. Ammar, A. Bourek, and A. Benakcha, ‘‘Robust SVM-direct torque con-
trol of induction motor based on sliding mode controller and sliding mode
observer,’’ Frontiers Energy, pp. 1–14, Jan. 2017, doi: 10.1007/s11708-
017-0444-z.
[30] A. Djoudi, S. Bacha, H. Iman-Eini, and T. Rekioua, ‘‘Sliding mode control
of DFIG powers in the case of unknown flux and rotor currents with XUDONG LIU received the B.S. and M.S. degrees
reduced switching frequency,’’ Int. J. Elect. Power Energy Syst., vol. 96, in automation from Qingdao University, Qingdao,
pp. 347–356, Mar. 2018, doi: 10.1016/j.ijepes.2017.10.009. China, in 2008 and 2011, respectively, and the
Ph.D. degree in power electronics and electric
drives from Shandong University, Jinan, China,
BINGKUN ZHAO received the B.Sc. degree in in 2016. Since 2016, he has been with Qingdao
electrical engineering and automation from Shan- University. His current research interests include
dong Jianzhu University, Jinan, China, in 2016. new energy systems, power electronics, and motor
He is currently pursuing the M.Sc. degree in con- drive.
trol science and engineering with Qingdao Univer-
sity, Qingdao, China. His research interests include
robot control, electrical energy conversion, and
motor control and applied nonlinear control.

HAISHENG YU received the B.S. degree in

electrical automation from the Harbin University HERONG WU received the B.Sc. degree in
of Civil Engineering and Architecture in 1985, automation from Qingdao University, Qingdao,
the M.S. degree in computer applications from China, in 1989. He is currently a senior Exper-
Tsinghua University in 1988, and the Ph.D. degree imentalist with Qingdao University. His current
in control science and engineering from Shandong research interests include computer control, pro-
University, China, in 2006. He is currently a Pro- grammable logic controller technology and appli-
fessor with the School of Automation Engineering, cation, process control, and motor control.
Qingdao University, China. His research interests
include electrical energy conversion and motor
control, applied nonlinear control, computer control, and intelligent systems.

VOLUME 6, 2018 43849