Automation in Construction 42 (2014) 100–111

Contents lists available at ScienceDirect

Automation in Construction
journal homepage: www.elsevier.com/locate/autcon

Dynamics of an overhead crane under a wind disturbance condition

Jerzy Tomczyk, Jacek Cink ⁎, Andrzej Kosucki
Division of Working Machines, Drives and Control, Lodz University of Technology, Poland

a r t i c l e i n f o a b s t r a c t

Article history: The following article discusses issues such as a crane control system with a state simulator, problems of load op-
Received 14 June 2012 eration and positioning under different wind disturbances, the main elements of a dynamic model with a state
Revised 19 February 2014 simulator, a method of wind disturbance compensation and the results of simulation tests.
Accepted 22 February 2014
© 2014 Elsevier B.V. All rights reserved.
Available online 17 March 2014

Keywords:
Automatic control
State simulator
Dynamics
Overhead crane
Simulation tests

1. Introduction An experimentally verified sway-free system for a boom crane with

disturbance observers and a trajectory generator which smoothes the
Most cranes carry loads suspended on ropes. The load is suspended reference trajectory online is presented in [17]. The whole control sys-
rigidly in a vertical direction and flexibly in a horizontal direction, tem is implemented in a real harbor mobile crane.
which causes load oscillations during working movement. Positioning A model of a crane with disturbances from sea-wave-induced ship
of the load in a horizontal plane requires an optimum control of mech- motion and ship-motion-induced container sway is presented in [15].
anisms which control the motion of the rope suspension point. The In [18], a control system of the crane carriage movement under a dis-
mechanisms must contain continuous speed control and position ad- turbance condition (e.g. from wind or friction forces) with an observer
justment systems. is presented. The system gives large deviations from the reference tra-
The conducted research included load sway limitation obtained by jectory during acceleration and deceleration with satisfactory position-
means of different methods of fuzzy control described in [6,9,10], as ing at the end of a cycle.
well as other methods using optimum or feedback control, e.g. [12,13]. References to papers that confirm the analogy of the motion of a load
In [7], the UMC platform and Fieldbus are used for crane and hoist control. suspended to a carriage to a mathematical pendulum, as well as the pro-
Most of the presented dynamic models make use of simplified over- posal of a sway damping method during deceleration from a constant
head crane laboratory models. Examples of such models are presented velocity motion to stop, based on the developed model of a pendulum
in [6,9,10]. Similarly some papers do not take into consideration phe- are presented in [19].
nomena existing in real objects, e.g. beveling or different parameters The problem of optimum control was solved by moving the load
of the pendulum for different directions of the bridge and carriage along a horizontal straight line and dumping load oscillations after the
movement. Examples of these are: [8,11,12]. starting and braking phases. A straight trajectory of the load was en-
Some of the articles refer to the laboratory models of rotary and sured due to the simultaneous operation of traveling and traversing
boom cranes. In [14] a dynamic model of a crane and methods of its mechanisms in the case of a bridge crane, and slewing and jib mecha-
open-loop control with load sway damping without the necessity of nisms in the case of a jib crane. These methods of load positioning
load inclination measurement is presented, but only through a proper were verified through simulation and experimental tests [2–4].
shape of the control function. The methods described above are not suitable for cranes working in
In [16] a crane control system with partial-state feedback is present- the open air. A force caused by the wind acting on a flexibly suspended
ed, with an integrator allowing for simultaneous positioning and load could change its determined trajectory. Our first method [5] of
damping of load sway. eliminating wind disturbances requires knowledge of wind force acting
on the load. To determine this factor, data such as the shape of the load
⁎ Corresponding author: Lodz University of Technology Division of Working Machines,
Drives and Control K-111 ul. Stefanowskiego 1/15, 90-924 Łódź, Poland.
or wind direction and velocity should be measured. This makes it possi-
E-mail addresses: [email protected] (J. Tomczyk), [email protected] (J. Cink), ble to calculate the correction of the load position which constitutes an
[email protected] (A. Kosucki). input for the mechanism adjustment system. The rope suspension point

http://dx.doi.org/10.1016/j.autcon.2014.02.013
0926-5805/© 2014 Elsevier B.V. All rights reserved.
 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111 101

vc v
S
vb

y carriage
LQ

x
bridge
mQ RQ

vQx H

β r yQ
Fig. 3. A physical model of overhead crane plane motion.
α
vQ v The symbols used in this model are described as follows:
Qy
C center of the overhead crane mass,
XQ
xl left co-ordinate of the center of the overhead crane mass,
xr right co-ordinate of the center of the overhead crane mass,
Fig. 1. The dynamic model of a rope-suspended load. Pl driving force of the left end carriage,
Pr driving force of the right end carriage,
Symbols used in Figs. 1 and 2: H horizontal component of the force in the rope,
S rope suspension point, Wl movement resistance force of the left end carriage,
vb velocity of point S along the direction of the bridge movement, Wr movement resistance force of the right end carriage,
vc velocity of point S along the direction of the carriage movement, mcl mass of the left end carriage,
v resultant velocity of the suspension point, mcr mass of the right end carriage,
LQ length of the ropes, mb mass of the bridge,
RQ force of wind action on the load, mc mass of the carriage,
H horizontal component of the force in the rope, Lec length of the end carriage,
xQ displacement of the load along the direction of the bridge movement, Lecw wheelbase of the end carriage,
yQ displacement of the load along the direction of the carriage movement, ll location of the carriage from the left,
vQx velocity of the load along the direction of the bridge movement, lr location of the carriage from the right,
vQy velocity of the load along the direction of the carriage movement, L overhead crane span,
vQ resultant velocity of the load, vcb velocity of the overhead crane mass center,
mQ mass of the load, ωb angular velocity around the overhead crane mass center.
r relative horizontal position of the rope suspension point relative to the load,
α angle defining the direction of r and H vectors,
β angle defining the direction of the wind pressure force vector on the load,
vw wind velocity,
same input signals. In the case of the state simulator, calculations
vwq relative velocity of the wind to the velocity of the load,
αw angle defining the direction of the wind velocity vector. are carried out for windless conditions, whereas in the case of the real
object the load suspended on ropes is under wind action. The action
of wind disturbances changes the motion of the real object (load)
should be moved along the corrected trajectory, so that the real load and the time waveforms of the load position are different than the
trajectory should be compatible with the assumed one. The drawback
of this method is the necessity of wind velocity and direction
measurement.
The method presented in this paper uses a state simulator [1]. One of
the elements of the crane automatic control system is a real object
whose mathematical model is called a state simulator, operating simul-
taneously in real time. The real object and the state simulator have the

β
α

β Fig. 4. The definition of points P and K and determination of the velocity of the suspension
point.
α
xp, start position of point S along the direction of the bridge movement,
xk end position of point S along the direction of the bridge movement,
yp start position of point S along the direction of the carriage movement,
yk end position of point S along the direction of the carriage movement,
vbmax maximum permissible speed of the bridge traveling mechanism,
Fig. 2. Relative air velocity with respect to the load. vcmax maximum permissible speed of the carriage traveling mechanism.
102 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111

Fig. 5. A diagram of the overhead crane control system with wind disturbance compensation.

theoretical ones (without disturbances). A deviation, i.e. the value of the The mathematical model of a load suspended on ropes in the space
difference between the theoretical and the real load position is entered of state variables taking into account wind action is as follows:
into the regulator as a correction signal. The regulator determines the
dvQx fQ fQ cQ
correction signal only for the real object control system as long as the ¼  v  cosðαw Þ− v þ x ð1Þ
deviation is limited to a minimal value. The presented method makes dt mQ wQ mQ Qx mQ Q
it possible to improve the accuracy of load movement without any
dvQy fQ fQ cQ
knowledge of wind speed and direction. ¼  v  sinðαw Þ− v þ y ð2Þ
dt mQ wQ mQ Qy mQ Q

dxQ
2. A dynamic model of an overhead crane and a control system ¼ −vQx þ vb −ðxl −ll Þ  ωb ð3Þ
dt
The dynamic model of a load suspended on ropes is presented in dyQ
Fig. 1. ¼ −vQy þ vc ð4Þ
dt
The determination of relative air velocity with respect to the load
vwQ is shown in Fig. 2. where vQx, vQy, xQ and yQ are the defined state variables.

a
b

Fig. 6. Determination of the load position deviation a) and of the corrected position of the load rope suspension point b).
 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111 103

The inputs of this model are components determining the velocity of overhead crane bridge and carriage plane motion. Examples of motion
the rope suspension point S: vb and vc and the angular velocity of the equations describing overhead crane dynamics in a general form are
bridge: ωb. as follows:
The state space method was used for a bridge and carriage dynamic
– a motion equation of the mass center:
description. A physical model of their elements was made. An example
of an overhead crane physical model is presented in Fig. 3.
Motion equations based on the assumption of rigid constraints with- dvb X
m ¼ F ð5Þ
out kinematic losses were used in the mathematical description of the dt

a) Compensation system awitched off

b) Compensation system awitched on

Fig. 7. Load oscillation damping: vw wind velocity, vsg determined velocity of the load suspension point, ΔQ deviation of the model (state simulator) relative to the real load
positions, ΔQSg load displacement relative to the determined load suspension point Sg, ΔS displacement of the determined load suspension point Sg relative to its real position S.
104 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111

– a rotational motion equation in a horizontal plane relative to the before a steady motion and after stopping at point K as well. The wind
mass center: disturbance compensation system operates continuously all the time,
independently of the control system.
dωb X The main part of the overhead control system with wind disturbance
I ¼ T: ð6Þ
dt compensation is presented in Fig. 5. Load motion in a horizontal plane is
possible owing to the overhead bridge crane traveling and traversing
The main task of the control system is to determine the control func- mechanisms. The input signals (Fig. 5) are as follows:
tion for moving the load from the initial point P (xp, yp) to the terminal
point K (xk, yk) (Fig. 4). The load rope suspension point S is moved along – determined velocity of the bridge traveling mechanism vbg,
the horizontal straight line PK and the load oscillations are damped – determined velocity of the carriage traveling mechanism vcg.

a) Compensation system switched off

b) Compensation system switched on

Fig. 8. Trajectories: Sg the given load suspension point trajectory, S the real load suspension point trajectory, Q the real load trajectory.
 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111 105

These signals are sent to the continuously operating electronic con- The determined xg and yg positions are compared with the real posi-
troller which generates the determined position of the bridge xg and tion of point S (coordinates x and y). Signals of the determined velocities
carriage yg, i.e. the determined coordinates of the rope suspension of the load rope suspension point: vbg, vcg, xg and yg are simultaneously
point S. Positions xg and yg are determined by the controller by integrat- transmitted to:
ing the determined signals of bridge velocity vbg and carriage velocity
a) a block which simulates the load suspended on ropes (the state
vcg in real time:
simulator),
b) a correction block of the bridge and carriage velocity and position.
Z Z
The state simulator solves the mathematical model of the load
xg ¼ vbg  dtyg ¼ vcg  dt: ð7Þ
suspended on ropes in real time. The model describes the load behavior

a) Compensation system switched off

b) Compensation system switched on

Fig. 9. Load oscillation damping: vw, vsg, ΔQ, ΔQSg and ΔS — as in Fig. 6.
106 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111

without wind action and determines the theoretical load position given The aim of the wind disturbance compensation system is to reduce
by the coordinates xQt and yQt. the load position deviation εQ to zero, which means that the load
The correction block, called the load position regulator, determines moves along the same trajectory as in the case of the absence of wind
the corrected signals of the load suspension point S, both in the direc- disturbances. The components of the load position deviation deter-
tion of the bridge (velocity vbgc, position xgc) and in the direction of mined by the measurement and control systems are transmitted as
the carriage relative to the bridge (velocity vcgc, position ygc). These sig- input signals to the position and velocity correction block of the load
nals depend on the load position deviation εQ whose components are rope suspension point (Fig. 5).
determined as follows: The method of determining the load position deviation error is
εQ x ¼ xQt −xQ εQ y ¼ yQt −yQ shown in Fig. 6a. A correction of the position of the load suspension
ð8Þ
point S causes the elimination of the load position deviation. The real
where xQ and yQ — the real load position components. load positions xQ and yQ are measured in real time.

a) Compensation system switched off

b) Compensation system switched on

Fig. 10. Trajectories: Sg the given load suspension point trajectory, S the real load suspension point trajectory, Q the real load trajectory.
 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111 107

The compensation system functions as a proportional regulator with 3. Overhead crane dynamics under a wind disturbance condition
a gain coefficient k, which determines the new position of the load sus-
pension point Sgc displaced relative to the theoretical position Sg by the Dynamic models of overhead crane drives and control were used for
vector ΔR in a steady motion: the development of computer programs to carry out laboratory over-
head crane simulation dynamic tests with disturbance. A comparison
of system dynamics can be made by switching on or off the compensa-
ΔR ¼ k  εQ εQ ¼ εQx þ εQy : ð9Þ
tion system in the model, which makes it possible to check the influence
of the state simulator on the overhead crane working movement dy-
The corrected velocity of point S is the same as the determined namics. The duty cycle covers load transportation from the starting
velocity (vsgc = vsg). point P to the end point K along the straight line PK with a simultaneous

a) Compensation system switched off

b) Compensation system switched on

Fig. 11. Load oscillation damping: vw, vsg, ΔQ, ΔQSg and ΔS — as in Fig. 6.
108 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111

motion of traveling and traversing mechanisms. Symbols x and y were Simulation tests including cases of the wind disturbance com-
assigned to represent the direction of the bridge movement and the di- pensation system switched on or off were carried out for compari-
rection of the carriage movement, respectively. The trajectory length PK son. In each case, load oscillation damping after the starting and
was determined so as to carry out three phases of movement: starting, braking phases was obtained for constant wind velocity v w =
steady motion and braking. 22 m/s and constant wind direction perpendicular to the line PK,
The main purpose of the research was to determine the influence of which is shown in Fig. 7. The load position deviation ΔQ is high
the control system with wind disturbance compensation on the over- when the compensation system is off and reduced to zero when
head crane dynamics with a state simulator. the system is on.

a) Compensation system switched off

b) Compensation system switched on

Fig. 12. Trajectories: Sg the given load suspension point trajectory, S the real load suspension point trajectory, Q the real load trajectory.
 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111 109

The load and its suspension point trajectory are important for the nominal hoisting speed vp = 10 [m/min]
evaluation of the overhead crane operation with wind disturbance. hoisting height H = 6.6 [m].
They are shown in Fig. 8 for the same wind.
The real trajectory of the load Q in Fig. 8a is displaced with regard to Experimental tests of the compensation system for an overhead
the determined trajectory PK. It is the result of wind action on the load crane are conducted only in the motionless state of the crane. The load
with the compensation system switched off. The load does not move hanging on ropes is displaced in a horizontal direction by a force
along the given trajectory in this case. When the compensation system which simulates wind pressure. The compensation system is turned
is switched on, the load is moved along the given trajectory PK (Fig. 8b) on and causes such a horizontal displacement of the rope suspension
and the trajectory of the load suspension point S is displaced with refer- point (opposite to the direction of the load displacement) that the
ence to the load trajectory, which compensates wind disturbance. load is held in the previous position by deflected ropes.
The results of research on wind velocity disturbance with a single Time-based charts that show the movements of the bridge and car-
impulse of wind velocity are shown in Figs. 9 and 10. The wind velocity riage driving mechanisms which compensate wind disturbances ap-
impulse has a shape of a half-sinusoid with an amplitude of 22 m/s and plied to the load are presented in Figs. 14–15.
time of activity of 2 s with the wind direction perpendicular to the line The results of experimental tests of the compensation system co-
PK. The impulse appears during steady motion. Load oscillations operating with the traveling mechanism (in the direction of the bridge
(Fig. 9a) cannot be damped when the compensation system is switched motion) are presented in Fig. 14. The bridge moves in a direction oppo-
off. The load oscillates after the braking phase as well (Fig. 9a). site to the direction of the disturbance force, which allows the load to re-
When switched on, the compensation system makes it possible to main at the start position.
stop the working movement of the load at the determined point K. In Fig. 15, in a similar manner as for the bridge, experimental tests of
Load oscillations are damped as well (Fig. 10b). the carriage are presented. In this case, the correct operation of the com-
The compensation system works properly with a linear increase of pensation system is evident and operational requirements are also met.
wind velocity from 0 to 22 m/s during 1 s in a steady motion and at a
wind direction perpendicular to the line PK (Figs. 11 and 12).
The reaction of the compensation system against a disturbance is 5. Conclusions
clearly visible in Fig. 11b. The end of a duty cycle with the load posi-
tioned in point K without oscillations is shown in Fig. 12b. Load oscilla- The conducted simulation tests confirmed the correct operation of
tion damping is impossible when the compensation system is switched the presented control system. In conditions of wind action, the position
off (Fig. 12a). of the rope suspension point was automatically changed to keep the
load being under wind pressure in the given position.
4. Experimental verification In all tests, wind velocity was assumed to be perpendicular to the
load trajectory, i.e. to be most unbeneficial. The maximum value of
Experimental tests were carried out on a real overhead crane (Fig. 13) wind velocity was 22 m/s, which resulted in wind pressure at the
characterized by the following parameters: level of 300 N/m2 according to the terminal permissible conditions of
crane operation.
load Q = 5 [t] Model research carried out for a constant wind direction and veloc-
bridge span L = 10 [m] ity (Figs. 7 and 8) showed that the load was moved along the assumed
bridge track length Lt = 16 [m] trajectory PK with high accuracy during working movement.
nominal bridge speed vm = 34 [m/min] The simulation tests conducted for a sinusoidal wind impulse which
nominal carriage speed vw = 34 [m/min] appeared in a steady motion confirmed the stability of the adjustment

Fig. 13. The experimental stand — an overhead crane.

110 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111

650 1300
- xmr [mm] W [N]
600 1200
x Qmr
550 Qmr [rad/5000]
1100

500 1000

450 900
ϕQmr
400 800

350 700

300 xmr 600

250 500

200 400

150 300

100 W 200

50 100
t [s]
0 0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105
-50 -100

-100
xQm -200

-150 -300

Fig. 14. Corrective motions of a “stopped” overhead crane bridge under disturbances acting on the load.

W disturbance force acting on the load perpendicularly to the bridge,

xmr corrective movement of the bridge,
xQm load position,
φQmr vertical angle of load sway.

system (Figs. 9 and 10). Although the adjustment time was equal to The results of the simulation tests carried out for a wind ramp signal
about 10 s, the oscillations of the load suspension point and load posi- increasing with time (from zero to 22 m/s) 1 s and appearing in a steady
tion decreased to zero (Fig. 9b), also after the braking phase. The devia- motion confirmed the stability of the adjustment system (Figs. 11 and
tion of the load position in relation to the determined trajectory PK was 12). The oscillations faded (Fig. 11b). The load was moved along the
low all the time (Fig. 10b). straight line PK with high accuracy (Fig. 12b).

650 1300
- xwr [mm] W [N]
600 x Qwr [mm] 1200

550 Qwr [rad/5000] 1100

500 1000
xwr
450 900

400 ϕQwr 800

350 700

300 600

250 500

200 400

150 300

100
W 200

50 100
t [s]
0 0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
-50 -100

-100 xQw -200

-150 -300

Fig. 15. Corrective motions of a “stopped” overhead crane carriage under disturbances acting on the load.

W disturbance force acting on the load parallel to the bridge

xwr corrective movement of the carriage,
xQw load position,
φQwr vertical angle of load sway.
 J. Tomczyk et al. / Automation in Construction 42 (2014) 100–111 111

The good quality of the wind disturbance compensation system was [7] N.A. Armstrong, P.R. Moore, A distributed control architecture for intelligent crane
automation, Autom. Constr. 3 (1) (May 1994) 45–53.
confirmed only on the grounds of simulation and simplified experimental [8] G. Boschetti, R. Caracciolo, D. Richiedei, A. Trevisani, Moving the suspended load of
tests. The satisfactory effects of theoretical and simplified experimental an overhead crane along a pre-specified path: a non-time based approach, Robot.
tests are the basis for wider experimental tests confirming good operation Comput. Integr. Manuf. 30 (2014) 256–264.
[9] Diantong Liu, Jianqiang Yi, Dongbin Zhao, Wei Wang, Adaptive sliding mode
of described system. The wider experimental tests of described overhead fuzzy control for a two-dimensional overhead crane, Mechatronics 15 (2005)
crane are presently carried. 505–522.
[10] Yu. Wen, Marco A. Moreno-Armendariz, Floriberto Ortiz Rodriguez, Stable adaptive
compensation with fuzzy CMAC for an overhead crane, Inf. Sci. 181 (2011)
References 4895–4907.
[11] Carlos Vázquez, Joaquin Collado, Leonid Fridman, Super Twisting Control of a Para-
[1] J. Tomczyk, et al., Badania jakości sterowania mechanizmami dźwignic z
metrically Excited Overhead Crane, Journal of the Franklin Institute, 2013. (Available
wykorzystaniem obserwatora stanu w warunkach zakłóceń działaniem wiatru
online 7 March 2013).
((Research on the Quality of Crane Mechanisms Control Using a State Simulator
[12] Ning Sun, Yongchun Fang, Xuebo Zhang, Energy coupling output feedback control of
Under Wind Disturbance Conditions) Praca badawcza Nr (Research work No.) 502
4-DOF underactuated cranes with saturated inputs, Automatica 49 (2013)
003), Politechnika Łódzka (Lodz University of Technology), Łódź, 2009.
1318–1325.
[2] J. Tomczyk, J. Cink, Automatyzacja transportu ładunków dźwignicami sterowanymi
[13] A.Z. AI-Garni, K.A.F. Moustafa, S.S.A.K. Javeed Nizami, Optimal control of overhead
numerycznie, suwnice — transport poziomy (Automation of load transport by nu-
cranes, Control Eng. Pract. vol. 3 (No. 9) (1995) 1277–1284.
merically controlled cranes, overhead cranes — horizontal transport), Przegląd
[14] Naoki Uchiyama, Huimin Ouyang, Shigenori Sano, Simple rotary crane dynamics
Mechaniczny (10/2001) 16–21.
modeling and open-loop control for residual load sway suppression by only hori-
[3] J. Tomczyk, J. Cink, A. Kosucki, Metody rozwiązywania problemów automatyzacji
zontal boom motion, Mechatronics 23 (2013) 1223–1236.
sterowania dźwignicami (Methods of solving problems of crane control automa-
[15] Keum-Shik Hong, Quang Hieu Ngo, Dynamics of the container crane on a mobile
tion), Przegląd Mechaniczny (1/2003) 9–16.
harbor, Ocean Eng. 53 (2012) 16–24.
[4] J. Cink, Sterowanie optymalne maszyn dźwigowych, (Optimum Control of Cranes)
[16] Naoki Uchiyama, Robust control of rotary crane by partial-state feedback with inte-
Zeszyty Naukowe Politechniki Łódzkiej, Łódź, 2001.
grator, Mechatronics 19 (2009) 1294–1302.
[5] J. Tomczyk, et al., Badania eksperymentalne i optymalizacja transportu ładunków
[17] Jorg Neupert, Eckhard Arnold, Klaus Schneider, Oliver Sawodny, Tracking and anti-
dźwignicami o ruchach obrotowych i wypadowych sterowanych za pomocą
sway control for boom cranes, Control. Eng. Pract. 18 (2010) 31–44.
mikroprocesorów ((Experimental Tests and Optimisation of Load Transport Using
[18] H. Aschemann, O. Sawodny, S. Lahres, E.P. Hofer, Disturbance estimation and com-
Cranes With Microprocessor-controlled Slewing and Jib Movements) Praca
pensation for trajectory control of an overhead crane, Proceedings of the American
badawcza Nr (Research work No.) 7 T07B 013 19), Politechnika Łódzka (Lodz Uni-
Control Conference Chicago, Illinois, June 2000, 2000, pp. 1027–1031.
versity of Technology), Łódź, 2002.
[19] U. BUGARIĆ, J. VUKOVIĆ, Optimal control of motion of the system based on mathe-
[6] Cheng-Yuan Chang, Kuo-Hung Chiang, Fuzzy projection control law and its applica-
matical pendulum with constant length, FME Trans. 30 (2002) 1–10
tion to the overhead crane, Mechatronics 18 (2008) 607–615.