Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract: Virtual simulation of the real behaviour of mobile harbour crane (MHC) without using the
traditional build-and-test method is an imperative approach to the design stage that can increase the
quality of the product by reducing manufacturing cost and errors. This paper introduces an engineering
model that describes the mechanical behaviour of MHC, and the control design for increasing the posi-
tion accuracy. Based on a concept of the MHC, a virtual mechanical model was created using SO-
LIDWOKS, which was then exported to the Automatic Dynamic Analysis of Mechanical System
(ADAMS) environment. This model was simulated to investigate the dynamic behaviour of the MHC
system. In addition, an adaptive siding mode PID controller was also developed in MATLAB/Simulink
to control the crane trolley position and suppress the swing angle of the load. This co-simulation de-
monstrates the reliability of the mechanical and control functionalities of the developed system.
Keywords: Adaptive sliding mode, mobile harbour crane, modeling and controlling, PID controller,
virtual prototype.
electrical solution is popularly used to anti-swing the 2. COMPOSITION AND WORKING PRINCIPLE
load. This solution was further classified into two kinds OF THE MHC SYSTEM
of control approach, open-loop control and closed-loop
control. The open-loop control scheme proposed in [4] The MHC is the crane system mounted on the floating
was not equipped with sensors. Its control objectives as shown in Fig. 2. It consists of a floating, frame system,
were based on predetermined speed control and support frame, trolley, spreader and boom. The floating
trajectory, which have been simulated to dispose of the can carry whole crane systems and the load. It works on
swing. This method is economical and stable with cranes the sea and is swayed by the wind and wave. The frame
that have a low natural frequency. However, it is system is a firm structure to withstand the total load of
inefficient with unstable plants, which are affected by the crane. It moves along the floating and can be adjusted
disturbances. Meanwhile, the closed-loop control horizontally to pick up a container. The support frame is
schemes were proposed as an anti-swing for the cranes. raised when the crane begins working and is lowered
These schemes are equipped with many types of sensors when stopped. This function aims to collapse the crane
to detect the sway angle of the load in linear/nonlinear system for convenient traveling. The trolley moves on
controls [5-11]. The swing angle signal is processed by the boom rail following the x-direction and is driven by a
observation and estimation models before being motor force. It drives the spreader to the desired position
transmitted to the controller. Several equipments were to pick and insert a container. The spreader is suspended
proposed to measure the swing angle of the load. on the trolley by a four-cable mechanism, and its
Yoshida et al. [10] proposed to use cameras as a non- function is to adjust the hooks of a container for lifting.
contact sensor for visual feedback control of the crane. In The clamping and positioning systems are used to locate
this case, a 3D camera was installed on the trolley to the container accurately.
measure the 3D position of the load. This approach is Due to working on the sea, the MHC is influenced by
effective for cranes that experience disturbances. the wind and wave. According to Jang et al. [14], the
However, this vision system is cost consuming and wind-induced drap force acting on the crane structures
difficult to maintain. In addition, longevity of vision over the seaway can be evaluated based on the
devices is gradually reduced when they are exposed to fundamental equation of drap force in aerodynamics:
the sea environment. In order to replace the vision
system, Kim et al. [11] introduced a new vision sensor- 1
FD (t ) = ρ CD AU 2 (t ) , (1)
less anti sway control system for container cranes, which 2
uses an inclinometer that is placed on the spreader to where ρ = Air density,
detect the sway angle of the load. However, the common CD = Drag coefficient,
inclinometers lack high precision and have slow response A = Projected area of a structure,
times. Therefore, an expensive inclinometer is needed to U(t) = Incident wind speed,
achieve high precision. To achieve more economical,
precise and quick responses, Park et al. [12] introduced a U (t ) = U (t ) + w(t ) . (2)
new approach, in which a tri-axial accelerometer was
used to estimate the swing. In this approach, the swing The Ū is a constant wind speed depending on the
angle is measured by the accelerometer based on the height above the sea level, and the w(t) is a randomly
difference of the fixed points between the trolley and the fluctuating turbulent wind speed. The wind-induced drag
spreader. Collaboratively, the above approaches were force, FD(t) can be written as follows:
aimed to detect the swing angle effectively and to design
a closed-loop control system to achieve an anti-swing of
the load. These approaches have achieved the desired
control objectives. However, the launch and deployment
of a new control product into a real work environment
require further time and cost for building, testing and
verifying on the physical model.
Hence, this paper proposes a virtual prototyping
simulation technology that integrates ADAMS and
MATLAB/Simulink. The ADAMS software is used to
create the virtual mechanical model, with the possibility
of virtual measurement of any parameters of any
components in the virtual model. MATLAB software is
well-known for designing a control system. Our co-
simulation model of both software has a merit in
simulating real behaviour of the mechanical system, and
implementing the closed-loop control of the whole
virtual prototype model. The simulations of results not 1. Floating; 2. Frame; 3. Support frame; 4. Boom;
only help designers to modify mechanical design but also 5. Trolley; 6. Spreader.
improve the control method. Fig. 2. The MHC modeling.
1206 Hong-Soek Park and Ngoc-Tran Le
Fig. 5. The virtual prototyping platform. Fig. 6. The Adams model of the MHC system.
1208 Hong-Soek Park and Ngoc-Tran Le
3 .5 ADAMS_uout
1
Trolley position [m]
3 .0 U To Workspace
Displacement
2 .5
1 Mux MSCSoftware Demux
2 .0
F_trolley
ADAMS Plant
1 .5
2
1 .0 theta
ADAMS_yout
0 .5
Mux Y To Workspace Demux
0 .0
0 10 20 30 40 50 ADAMS_tout
T im e [sec] Clock T To Workspace
Fig. 7. The comparison between simulation and experi- Fig. 10. The ADAMS block in adams_sys.
ment responses of the trolley displacement.
the input and output variables are firstly defined in the
0.10
Sim ulation ADAMS model. The input signal is the force that
0.08 E xperim ent
controls the trolley movement. Meanwhile, the output
0.06
signals are the measured parameters of the trolley
Longitudinal sway [rad]
0.04
position and the swing angle of the load, respectively.
0.02 Subsequently, this model is exported to MATLAB/
0.00 Simulink. Fig. 9 shows the connection between Adams
-0.02 and Matlab in the co-simulation model.
-0.04 In MATLAB environment, a .m file and an adams_sys
-0.06 are created. The adams_sys presents the non-linear
-0.08 MSC.ADAMS model (i.e., the MHC mechanical system)
-0.10 as shown in Fig. 10. The ADAMS block is created based
0 10 20 30 40 50 on the information from the .m file [17].
T im e [sec] This adams_sys is used to build a control system in
Fig. 8. The comparison between simulation and experi- MATLAB/Simulink in which the input signal is the
ment responses of the sway angle of load. generated signal from the controller, and the output
signals are the measured values of the trolley
Therefore, the ADAMS model can be used to imitate the displacement and swing angle of the load, respectively.
characteristics of the MHC mechanical model. This
model also can be used to simulate, analyse, test, and 4.2. Controller design
validate the behaviour of the mechanical and the control As analysed above, the MHC model is a nonlinear
systems. system that is operated in harsh environments with
external disturbance and system parameter variations. In
4. DEVELOPING A CONTROL SYSTEM FOR this condition, a traditional PID controller, with three
THE MHC MECHANISM PID gains (proportional gain KP, integral gain KI, and
derivative gain KD) are fixed, cannot be applied for
4.1. Creating an adams_sys in MATLAB/Simulink controlling the position and angle of the MHC system.
Constructing a control system for the virtual MHC In order to achieve system robustness against param-
model is necessary for co-simulation of the two separate eter variations, this paper proposes an adaptive sliding
simulation programs into a whole system. The control mode PID controller (ASMP), which combines the
design is developed based on ADAMS/Control and advantages of PID and sliding mode controllers [18]. Fig.
MATLAB/Simulink. To export the virtual MHC mech- 11 shows the block diagram of the adaptive sliding mode
anical model from ADAMS to MATLAB environment, PID control system for the MHC.
Modeling and Controlling the Mobile Harbour Crane System with Virtual Prototyping Technology 1209
]
⎧1 if sx > 0 [m 1.5
no
⎪ iit
sgn( sx ) = ⎨ 0 if sx = 0 (18) so
p
⎪−1 if s < 0 ye 1.0
ll
⎩ x or
T
Equation (14) proves that the sliding surface is stability. 0.5
The control input for controlling the MHC model is
composed of the trolley position control input ux and the 0.0
angle control input uθ as: 0 5 10 15 20 25 30
Tim e [sec]
u = u x + uθ . (19) (a) Trolley motion.
1.0
A SM P control
0.8 N o control
5. SIMULATION RESULTS
0.6
0.4
In this section, the simulation is implemented in the ]
-0.2
y
disturbances are changed by the height and frequency of a
w
S
2 .5 2 .5
] ]
[m 1 .5 [m
no no 1 .5
iit iit
so so
p p
eyll 1 .0 ye
llo 1 .0
roT rT
0 .5 0 .5
0 .0 0 .0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T im e [sec] T im e [sec]
(a) Trolley motion. (a) Trolley motion.
1 .0 1 .0
A S M P c on trol A S M P con trol
0 .8 N o c on trol 0 .8 N o control
0 .6 0 .6
0 .4 0 .4
] ]
0 .2
d
0 .2
d
a a
r r
[ [
e e
l
0 .0
l
g
n 0 .0 g
n
a
a
y
-0 .2
y
a
w
-0 .2 a
w
S S
-0 .4 -0 .4
-0 .6 -0 .6
-0 .8 -0 .8
-1 .0 -1 .0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T im e [sec] T im e [sec]
(b) Sway motion. (b) Sway motion.
Fig. 13. The position and angle responses of ASMP Fig. 15. The position and angle responses of ASMP
control with hw = 0.04 m, fw = 1.5 rad/s, l = 1.2 control with hw = 0.04 m, fw = 3 rad/s, l = 1.2 m,
m, ml = 148 kg. ml = 148kg.
2 .5 2.5
A S M P c on trol A S M P control
N o c on trol N o control
2 .0 2.0
] ]
[m 1 .5 [m 1.5
no no
iit iit
so so
p p
ye 1 .0 eyll 1.0
llo
rT roT
0 .5 0.5
0 .0 0.0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T im e [se c ] T im e [sec]
(a) Trolley motion. (a) Trolley motion.
1 .0 1 .0
A S M P con trol A S M P con trol
0 .8 N o con trol 0 .8 N o con trol
0 .6 0 .6
0 .4 0 .4
] ]
0 .2 0 .2
d d
a a
r r
[ [
e e
l
g
n 0 .0 l
g
n
0 .0
a a
y
a
w
-0 .2 y
a
w
-0 .2
S S
-0 .4 -0 .4
-0 .6 -0 .6
-0 .8 -0 .8
-1 .0 -1 .0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T im e [sec] T im e [sec]
(b) Sway motion. (b) Sway motion.
Fig. 14. The position and angle responses of ASMP Fig. 16. The position and angle responses of ASMP
control with hw = 0.02 m, fw = 3 rad/s, l = 1.2 m, control with hw = 0.02 m, fw = 1.5 rad/s, l = 1.5
ml = 148 kg. m, ml = 350 kg.
1212 Hong-Soek Park and Ngoc-Tran Le
2 .5 2 .5
A S M P c on trol A S M P c on trol
N o c on trol N o con trol
2 .0 2 .0
] ]
m
[ 1 .5 [m 1 .5
no no
tii tii
so so
p p
eyll 1 .0 eyll 1 .0
roT roT
0 .5 0 .5
0 .0 0 .0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T im e [sec] T im e [se c ]
(a) Trolley motion. (a) Trolley motion.
1 .0 1 .0
A S M P c on trol A S M P c on trol
0 .8 N o con trol 0 .8 N o c on trol
0 .6 0 .6
0 .4 0 .4
] ]
d
a
r
[
0 .2 d
a
r
[
0 .2
e e
0 .0
l
0 .0
l
g g
n n
a a
-0 .2
y
-0 .2
y
a a
w w
S S
-0 .4 -0 .4
-0 .6 -0 .6
-0 .8 -0 .8
-1 .0 -1 .0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T im e [sec] T im e [sec]
(b) Sway motion. (b) Sway motion.
Fig. 17. The position and angle responses of ASMP Fig. 19. The position and angle responses of ASMP
control with hw = 0.04 m, fw = 1.5 rad/s, l = 1.5 control with hw = 0.04 m, fw = 3 rad/s, l = 1.5 m,
m, ml = 350 kg. ml = 350 kg.
2 .5
A S M P c o n tro l Through the comparison of the position and angle
N o c o n tro l
responses between no control and proposed ASMP
2 .0
control approach as shown in Figs. 12-19, the achieved
] performances on the virtual prototyping model are
[m 1 .5
no evaluated based on the above criteria. The trolley
iit
so
p
ye
position performances are evaluated in terms of settling
llo 1 .0 time, overshoot, and error. Meanwhile, the sway angle
rT
performances are evaluated based on sway amplitude and
0 .5 its settling time.
Tables 3, 4, 5, and 6 show the comparison of both
0 .0 positioning and angle performances between no control
0 5 10 15 20 25 30
T im e [sec ] and proposed ASMP control approach in all cases of
(a) Trolley motion. disturbance and system parameter variations.
1 .0 The evaluating for the positioning and angle
A S M P con trol
0 .8 N o con trol performances is based on comparison detail in Tables.
0 .6 In all simulation cases, the overshoot appears less than
0 .4 2%, settling time maintains in range 4-5 second, and
]
d
a
r
[
0 .2 steady state error approximately zero. The positioning
e
l
g
n
0 .0 response is tracking the desired position, and the
a
y
a
w
-0 .2
S