5

Three-Dimensional Crane Modelling and Control

Using Euler-Lagrange State-Space Approach and
Anti-Swing Fuzzy Logic
Andrei Aksjonov (Doctoral student, Tallinn University of Technology), Valery Vodovozov (Professor, Tallinn
University of Technology), Eduard Petlenkov (Associate professor, Tallinn University of Technology)

Abstract The mathematical model of the three-dimensional system engineering. The main problems still remain in the
crane using the Euler-Lagrange approach is derived. A state- three-dimensional crane accurate positioning and in the load
space representation of the derived model is proposed and
explored in the Simulink environment and on the laboratory
swing elimination during the motion.
stand. The obtained control design was simulated, analyzed and To control the crane speed and positioning, many different
compared with existing encoder-based system provided by the controllers have been offered, including the classical
three-dimensional (3D) Crane manufacturer Inteco . As well, an proportional-integral (PI), proportional-differential (PD), and
anti-swing fuzzy logic control has been developed, simulated, and proportional-integral-differential (PID) controllers, linear-
analyzed. Obtained control algorithm is compared with the quadratic regulators, model predictive control systems,
existing anti-swing proportional-integral controller designed by
intelligent fuzzy logic controllers (FLC), and many others. All
the 3D crane manufacturer Inteco . 5-degree of freedom (5DOF)
control schemes are designed, examined and compared with the the controllers have their own advantages and disadvantages.
various load masses. The topicality of the problem is due to the At present, the intelligent control schemes built on the fuzzy
wide usage of gantry cranes in industry. The solution is proposed logic are widely distributed in automation control engineering.
for the future research in sensorless and intelligent control of Mainly, they are the most suitable and beneficial in both
complex motor driven application. multi-input, multi-output (MIMO) and multi-input, single-
output (MISO) systems. They have approved their success in
Keywords Control systems, Lagrangian functions,
multidimensional systems, fuzzy control, variable speed drives. process control in many areas such as flight, engine, position
and process controls, control of vehicle, navigation, load
I. INTRODUCTION estimation and many others.
Substantially, some interesting ideas were proposed to
Nowadays, the three-dimensional (3D) cranes also known replace the encoders with the model-based controllers making
as overhead gantry cranes are widely used in industry. the control safer, cheaper and more accurate due to the fact
Particularly, for vessel loading and unloading, to handle the that encoders provide and accumulate errors. Furthermore, it is
various masses and dimensions, lift containers, etc. They are important to design the control logic stable and adaptive to
applied in power industry, automotive lines, ship and airplane various masses of the load.
building and other heavy manufacturing. The modeling of the control systems is one of the most
Contemporary cranes are operated mostly manually. This important parts in control engineering. It helps to understand
makes the system dangerous because of the numerous both physical and theoretical behaviour of the system. Finally,
accidents and accuracy loss. The acceleration or deceleration it helps to choose and design the appropriate control algorithm
of the crane may result in undesirable load swing causing based on the needs. In this context, researchers from different
human accidents or damage the crane or the load. To control universities around the world have shown their solutions in
the swing, the operators need to be experienced in the crane modeling and building algorithms for three-dimensional crane
maneuvering. New algorithms used in automated systems may system described in [1] - [4]. In [2] is also used the Lagrange
help the operator to control the load swing or can even replace theory. The Cartesian coordinate robot is described in [5]. As
the operator. well as Inteco proposed in [6] original mathematical model
Today, the cranes are being built with semi- or full- derivation. However, none of the researchers were found
automation capability. Many studies have been conducted in where the state-space approach is applied to build the control
order to find the best, cheapest, and optimal solution to control system for the overhead crane.
cranes. Researches in the area are intended to improve the In [7] and [8] the fuzzy logic was used to design the control
accuracy and to minimize the system human interface system and in [9] the same approach was applied to provide an
resulting in the increasing work efficiency provided by the anti-swing control. Fuzzy logic control for load swing
cranes, avoiding accidents, and minimizing dependence on the elimination performance in [10] guarantees both accurate
system operator by involving automation control. positioning and prompt damping of load swing. Moreover, the
Thanks to diminishing of the human factor risk, the system author provides the 3D Crane system modelling as well.
may achieve precise control and improve load positioning. Authors of [11] presented the separate fuzzy logic
Thus, the control problem plays an important role in crane controllers for the 3D Crane trolley and hoisting. The anti-

DOI: 10.1515/ecce-2015-0006
 6

swing FLC has been also offered and combined with the
trolley control. Mathematical modeling of the overhead gantry
crane based on the single pendulum system, along with the
digital simulation of the fuzzy logic was shown in [12] and
[13]. In the latter, the fuzzy logic control has been compared
with the Linear Quadratic controller. In [14] another
proportional-derivative (PD)-type fuzzy controller has been
designed as an anti-swing algorithm. Nevertheless, the results
show only the horizontal position of the cart.
In [15] and [16] an adaptive fuzzy controller for 3D Crane
was designed. The results were simulated with varying length
of flexible-wire of the load as well with different masses.
Unfortunately, the results were guided for the only motion
between the two points: initial and final. No information has
been provided for the behavior of the overhead gantry crane Fig. 1. Laboratory 3D Crane setup.
system with constantly varying length of the lift-line and
motion between positions. In this case the oscillation of the
payload may not be able to dissolve before the next position
motion and the oscillation will remain constant or accumulate
the amplitude.
Other anti-swing solutions based on fuzzy logic regulator
were proposed in [17] - [20]. The last paper also showed the
PID regulator simulations for the 2-position displacement.
Finally, the author in [21] theoretically described the
difference between PID and fuzzy controllers.
The current paper represents the results obtained by the
authors in the mathematical modelling using Euler-Lagrange
equation and state-space approach. An anti-swing fuzzy logic
control design of the 3D Crane is proposed. Different masses
are given to the overhead gantry crane laboratory setup while (a) (b)
the FLC experiments conducted. The results are analyzed and Fig. 2. 3D Crane: (a) workspace (b) schematic drawing.
compared in the similar experimental conditions tested on the
TABLE I
existing PI controller provided by the 3D Crane manufacturer
3D CRANE DYNAMIC PARAMETERS DESCRIPTION
Inteco.
The main tasks of the presented work are as follows: Symbol Description
to derive the mathematical model for the 3D crane mc Mass of the cart; mc =1.16 kg
ml Mass of the payload; ml =0.58 kg
to simulate the system using Simulink environment and
mr Mass of the moving rail; mr =2.2 kg
to compare it with the existing encoder-based system provided
xp X-axis position of the payload
by the crane manufacturer
yp Y-axis position of the payload
to tune the derived controller as close as possible to the zp Z-axis position of the payload
existing one in order to be able to replace the encoder-based g Gravitational constant
system with the model-based unit Fx Force driving the rail with the cart
to design a fuzzy logic payload anti-swing 5DOF control Fy Force driving the cart along the rail
for 3D Crane accurate positioning l Length of the lifted load; l(max)=820 mm
to simulate and compare the results for FLC and PI Tx X-axis tension force
controllers with different payload masses and varying length Ty Y-axis tension force
of the lifted load. S Reaction force acting while lift the load
The paper is organized in the following way. First, the 3D xw X-axis position of the cart in workspace; x w(max)=920 mm
yw Y-axis position of the cart in workspace yw(max)=920 mm
Crane experimental setup is described. Second, the model
Angle between the lift- line and its projection on XY
derivation is explained. Then, the major experimental results
Angle between the negative direction of the Z-axis and
are discussed and solutions are drawn. Fourth, the anti-swing
projection of the lift- line on XY plane
5DOF fuzzy control is designed. Next, the anti-swing 5DOF
PI and FLC regulators are simulated at various payload masses
II. 3D CRANE EXPERIMENTAL SETUP
and changing load lifting-line. Finally, the simulation and
experimental results are compared. The research was conducted on the computer-controlled
three-dimensional crane setup 3D Crane from the

DOI: 10.1515/ecce-2015-0006
 7

manufacturer Inteco (Fig. 1) representing the small copy of The variable q relates to a generalized coordinate for each
an industrial crane [6]. output. The final result of the dynamic system for the 3D
3D Crane control system is suitable for testing complex Crane is represented by 5 nonlinear second-order differential
nonlinear MIMO control algorithms. The crane is driven by equations.
three direct current (DC) variable-speed drives. It consists of a B. State-Space Representation
payload hanging on a pendulum-like lift-line with a motor The dynamic model obtained is highly nonlinear
mounted on a cart [6]. The crane operation feedback is multivariable system. In order to design the control for the 3D
fulfilled by the Hall-effect switches used as the encoders. Crane, the state-space representation is used as the best
Switches measure the positions and the angles of the payload instrument to highly study complicated physical systems.
in real-time condition. However, before the control design, the system dynamical
The software operates under MS Windows NT using equations may be subjected to the approximations. Assuming
MATLAB /Simulink environment and Real-Time Window the small deviation of the payload along the z-axis executed by
Target (RTWT) toolbox package [6]. The control aims to the fast controller, the dynamic model can be simplified as
provide the signals to adjust three DC motors, moving along follows:
the X, Y and Z-axes. sin (9)
The mounting frame makes a support and a flexible cos 1
construction of the system. The construction is rigid due to the
reason it is fixed on the floor. The input forces for the 3D Crane are stated as the
The schematic drawing of the 3D Crane and an appropriate following vector:
workspace are presented in Fig. 2. All the presented dynamic
parameters in the Fig. 2 are described in Table 1. There are 5
u u xu yu z T
(10)
parameters measured by the encoders: xw, yw, l, and . These The state-space operational point is stated as follows:
parameters serve also as the outputs of the control system.
Mention that the positions of the cart and the payload are

x 0 x 0 x 0 y 0 y 00 0 0 0l 0l0
(11)
T

The A, B, C and D matrices are calculated taking into

restricted by the crane workspace and lifting-line of the load.
account the steady-state operation point x 0=0 and input forces
The dimensions of the limits are also presented in Table 1. vector u=0. These approximations are adequate due to the rule
where all the derivatives of the state variables, measured
III. MODEL DESCRIPTION AND SIMULATION
variables, and input forces must be equal to zero when the
A. Euler-Lagrange Dynamic Model Derivation expected steady-state position of the cart is obtained.
Finally, the state equations are given by
First, define the mathematical representation of the payload
as follows: x1 xw x2
xw x (1) 1 ml g
x2 xw ux
mc mr ml mc mr ml
yw y (2)
x3 y w x4
x p x l cos sin (3)
1 ml g
y p y l sin (4) x4 yw uy
mc ml mc ml
z p l cos cos (5)
x5 x6
The dynamic representation of the model is derived using (12)
1 g mc mr 2ml
the Lagrangian approach. The equation is defined as the x6 ux
difference of kinetic energy of the system K and its potential l mc mr ml l mc mr ml
energy P. In our case, K is the kinetic energy of both the crane
x7 x8
and its load whereas P defines the potential energy of the load:
1 g mc 2ml
1 x8 uy
L K P mv 2 mgl (6) l mc ml l mc ml
2
The speed of the load is defined as: x9 z w x10
1
v 2 x 2p y 2p z 2p (7) x10 zw uz
2ml
The Euler-Lagrange formulation of the dynamic system:

d L L All the differential equations for the state-space A, B, C and

Fi (8) D matrix elements have been calculated using the MATLAB
dt q i qi operation diff.
where i = 1, 2, 3, 4, 5

DOI: 10.1515/ecce-2015-0006
 8

1
FLC
2 1 3
2

Defuzzification
3 Inference

Fuzzification
r(t) mechanism
u(t) y(t)
Plant

Rule-base
(a) (b)

1 2
Fig. 4. Fuzzy logic controller block diagram: FLC fuzzy logic controller, r(t)
3 reference input, u(t) process inputs, y(t) process.

x
rx
ex
gex ux
(c) FLC gux
d/dt gax
X
g
2 1
2
y
ry
ey x
gey
uy 3D y
1 d/dt gay FLC guy
Crane
Y
g
z
z
(d) (e)
rz
Fig. 3. The 3D Crane state-space representation and Inteco model simulation ez gez uz
results: (a) X-axis output, (b) Y-axis output, (c) Z-axis output, (d) angle of FLC guz
payload output, (e) angle of payload output. In (a) (c): 1 input force, 2 d/dt gaz Z
state-space representation and 3 Inteco model. In (d) and (e): 1 the output
of Inteco model and 2 the state-space output.
Fig. 5. 3D Crane fuzzy logic control system: rx, r y and rz reference X, Y and
Z-axes positions; x, y and z real X, Y and Z-axes positions, X-axis angle;
C. Simulation Y-axis angle; e x, ey and ez X, Y and Z position errors; a x, ay and az
change in X, Y and Z position errors; u x, u y and uz X, Y and Z process inputs;
Both the designed state-space and Inteco models were g scaling gain for appropriate term; FLC X, FLC Y and FLC Z X, Y and Z
located in the same Simulink workspace window and the fuzzy logic controllers.
same input signals were applied to both models. Three traces
are added to view the input, the Inteco output of the model NL NS Z PS PL
and the output of the state-space representation. Therewith, the
additional traces are placed to observe and study the outputs of
the payload angles what cause the payload oscillation
problem.
As the experimental results did not completely match the
state-space representation output, the correction gain has been
-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1
found to scale axes positions for the state-space approach. The
output position respond is easy to tune by simple dividing the (a)
output steady-state value to the input force value. The gain =
[1.33 0.8 0.5] T was chosen. NL NS Z PS PL
The simulation results are presented in Fig. 3. The positions
along the X, Y and Z-axes were designed to follow the cart and
the payload lift-line according to the input force. The control
logic has been also tuned according to the design.
Analysis shows that there exist some error between the
encoder outputs and the state-space representation of the
-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1
model derived using the Euler-Lagrange approach.
Particularly, the amplitudes are not the same comparing to the (b)
3D Crane model and state-space representation, though the
Fig. 6. Membership function plot for the fuzzy logic controller inputs: NS
frequencies of the oscillation for both and angels are very negative small, NL negative large, Z zero, PS positive small, PL
close to each other. positive large: (a) input triangular membership function, (b) output Gaussian
membership function.

DOI: 10.1515/ecce-2015-0006
 9

In general, there can be several reasons explaining, why the the tables, only few linguistic sentences are presented here,
encoders and the state-space representation do not match each because the 3D Crane system requires more complex logic
other in amplitude. rules with 3 inputs and 1 output.
One of the reasons is that many dynamics such as viscose In total, there are 125 modus ponens (If-Then) form rules
friction coefficient, motors and encoder dynamics have been for both X and Y-axes. For Z-axis there are 25 modus ponens
neglected while modeling the system. Moreover, the form rules due to excluded angle oscillation output. The
disturbances that present in real-world environment are not examples of the linguistic rules are as follows:
considered. Also, some other components may influence the If error is NL and change-in-error NS and angle PS
system. Then force is NL
Next reason is that the 3D Crane frame can be tilted. Finally If error is Z and change-in-error Z and angle PS
not trivial, the encoder errors may appear. Of course, many Then force is PS
other features can trouble the correct dynamics. If error is NS and change-in-error ZE and angle PL
Then force is PS
IV. FUZZY LOGIC CONTROLLER The logic is as follows: if the motion force is very high in
A block diagram of fuzzy logic controller is presented in positive direction and the angle deviation is very large in
Fig. 4. The fuzzy system is static nonlinear mapping between negative direction, then apply force in opposite direction.
Precise explanation of how do the decisions in membership
inputs and outputs [22].
functions are made is described in [21] and [22].
The inputs and outputs of the system are crisp meaning
that they are real number. The fuzzification block converts the
B. Tuning the FLC via Scaling Universes of Discourse
crisp inputs to fuzzy sets. The inference mechanism uses the
fuzzy rules in the rule-base to produce fuzzy conclusion. The In order to scale the universes of discourse, the classical
defuzzification block converts these fuzzy conclusions into control engineering method is used. The tuning gains are
crisp outputs [22]. added to all the inputs and outputs of the fuzzy logic
controller. However, it is important to remember that tuning
A. Controller Design
the scaling gains is not the only solution for fuzzy logic
The MATLAB Fuzzy logic Toolbox is used to design the controller. The system can be also tuned via membership
FLC. First of all, it is important to define the crisp inputs. As it function shapes, positioning and number and type of rules
has been mentioned before, the system has three input forces [22].
to drive the crane along the X, Y and Z-axes. There are five As shown in the control logic system (Fig. 5), the scaling
output parameters: x, y, z, and . gains gex, gey, gez, gax, gay and gaz are added to both the
According to the scheme in Fig. 4 the outputs of the process proportional and the derivative terms of each of the controlled
are used as the controller inputs. We will therefore have axis X, Y and Z. The gains gux , guy and guz are also placed
MIMO system with 8 inputs and 3 outputs. For simplicity, the between the FLC and the 3D Crane. Further as we are
regulation of the X, Y and Z-axes will be divided separately considering the information about and angles of oscillation
onto MISO system. as another crisp input and their universes of discourse are not
It is important to note, that only one MIMO controller for known, the gains g and g are also introduced.
all the axes may be designed. At that, the system becomes All the mentioned gains are found during the experiment
more complex without any performance improvement. simulation. For the accurate positioning the standard solution
To develop the anti-swing regulator, the FLC also needs the was used: decrease the change-in-error term gain and barely
feedback from the angles and oscillation. X and Y-axes increase the error loop gain. On the whole, the scaling gains
therefore require three inputs and one driving force output for are chosen as follows: gex = 2, gey = 2, gez = 0.1, gax = 0.002,
the 3D Crane motors. gay = 0.04, gaz = 5, gux = 2, guy = 5, guz = 40, g = 2 and g = 2.
As soon as all the inputs and outputs are defined, the control
system can be specified. The block diagram of the designed V. EXPERIMENTATION R ESULTS
FLC is presented in Fig. 5. The simulation positioning results for various payload
The next step is to plot the membership functions of the masses of the PI anti-swing 5DOF control designed by the 3D
controller. The membership function quantifies the meaning of Crane manufacturer Inteco are provided in Fig. 7 Fig. 9.
the inputs and outputs. It can be seen in Fig. 6 that for the The payload oscillation results for both and angles are
crisp inputs the triangular membership function is used, shown in Fig. 10 and Fig. 11.
whereas the Gaussian membership functions are used for the The simulation positioning results for various payload
outputs. The universes of discourse L for inputs and outputs masses of the anti-swing 5DOF fuzzy logic control designed
are normalized on the range [1, 1]. The system is thus tuned in this paper are provided in Fig. 12 Fig. 14. The payload
via the scaling universes of discourse with the gains obtained oscillation results for both and angles are shown in Fig. 15
in experiments. and Fig. 16.
Finally, the design is concluded with the fuzzy linguistic For the fare comparison the same input forces have been
rules. Here it is important to imagine the behavior of the 3D given to the X, Y and Z-axes during the simulations. In
Crane system. Despite that the rules are usually presented in addition, before every simulation the object have been placed

DOI: 10.1515/ecce-2015-0006
 10

to Home position, where all the axes locations are equal to 0

as well as the encoders are reset to initial values.
Various payload masses have been given to the 3D Crane in
order to observe the differences and adaptation of the
controllers to the different conditions. Payload masses as well
as the reference position lines colors are shown in the box for
every plot in Fig. 7 Fig. 16.
Although, the respond of the Z-axis for both anti-swing PI
and fuzzy logic controllers are quite similar for every load
mass (Figs. 9 and 14), the controllers have significant
differences. Z-axis control is necessary to study the impact on
the regulators positioning and anti-swing control with
changing payload lifting-line.
It can be clearly seen from Figs. 7 and 8 that the positioning
in X and Y-axes for PI controller is very accurate. However, it
can be viewed the endpoint disturbance. The behavior is due
to controller reaction on the payload oscillation angle. On one Fig. 7. Anti-swing 5DOF PI regulator X-axis position simulation results.
hand, PI controller has fast respond to positioning. On another
hand, the oscillation of the load may reach up to 10 degrees,
what makes the cart and the rail bounce near the reference
position in order to eliminate the swing.
The oscillation angles (Fig. 11) and (Fig. 10) for anti-
swing PI regulator are very large. Not to mention, the angles
cannot be stabilized before the next motion. The payload
remains with the slight oscillation before another pulse.
Fuzzy logic controller, on the contrary, is not showing the
precise positioning, what can be viewed in Fig. 12 and Fig. 13.
The respond is quite fast, though it becomes slower getting
close to the reference point. The reason behind it may be the
fuzzy rule-based decision: when positioning is around the
reference point, the force applied starts to diminish slowly.
The positioning error in this case is approximately 0.025 m for
both X and Y-axes, what can cause a problem for bigger
dimension system.
Nevertheless, the designed anti-swing fuzzy logic controller
assimilates the payload mass change very well. Moreover, the Fig. 8. Anti-swing 5DOF PI regulator Y-axis position simulation results.
elimination of the oscillation of the payload (Figs. 15 and 16)
is much better comparing to the PI regulator what appears to
be the goal of the designed control algorithm.
Notwithstanding the positioning error, both (Fig. 16) and
(Fig. 15) angles can be stabilized around 0 degree before the
next motion. Also the swing index does not exceed 6 degrees
for and 8 degrees for while in PI regulator both angles are
higher. Again, neither the influence of the various payload
masses nor the lifting-line changes have been descried to
impact on the payload swing elimination.

Fig. 9. Anti-swing 5DOF PI regulator Z-axis position simulation results.

DOI: 10.1515/ecce-2015-0006
 11

Fig. 10. Anti-swing 5DOF PI regulator angle of the payload oscillation

Fig. 13. Anti-swing 5DOF fuzzy logic regulator Y-axis position simulation
simulation results.
results.

Fig. 11. Anti-swing 5DOF PI regulator angle of the payload oscillation Fig. 14. Anti-swing 5DOF fuzzy logic regulator Z-axis position simulation
results.
simulation results.

Fig. 15. Anti-swing 5DOF fuzzy logic regulator angle of the payload
Fig. 12. Anti-swing 5DOF fuzzy logic regulator X-axis position simulation oscillation simulation results.
results.

DOI: 10.1515/ecce-2015-0006
 12

[3] M. Z. Mohd Tumari, M. S. Saealal, M. R. Ghazali and Y. Abdul Wahab,

H controller with graphical LMI region profile for Gantry Crane
System, The 18th International Symposium on Artificial Life and
Robotics (AROB 18th 13), Daejeon, Korea, 2013, pp. 346 351.
[4] H.-H. Lee, Modeling and Control of a Three-Dimensional Overhead
Crane, Journal of Dynamic Systems, Measurement, and Control,
Vol. 120, pp. 471476, 1998.
[5] M. Callegari, F. Cannella, S. Monti, C. Santolini, Dynamic Models for
the Re-Engineering of a High-Speed Cartesian Robot, 2001
IEEE/ASME International Conference on Advanced Intelligent
Mechatronics, Como, Italy, 2001, pp 560565.
[6] Inteco Limited 3D Crane Users Manual: MATLAB R2009a/b,
R2010a/b, R2011a/b PCI version, Available at:
http://www.inteco.com.pl/products/3d-crane/ [Accessed: Feb. 20, 2015].
[7] P. Petrehus, Zs. Lendek and P. Raica, Fuzzy modeling and design for a
3D Crane, 3 rd IFAC International Conference on Intelligent Control and
Automation Science, Chengdu, China, 2013, pp 479484.
[8] D. M. Trajkovi1, D. S. Anti, S. S. Nikoli, S. Lj. Peri and M. B.
Milovanovi, Fuzzy Logic-Based Control of Three-Dimensional Crane
System, Automatic Control and Robotics Vol. 12, No. 1, pp. 31 42,
2013.
[9] Z. Jovanovi, A. Peri, S Nikoli, M. Milojkovi and M. Milosevi,
Fig. 16. Anti-swing 5DOF fuzzy logic regulator angle of the payload Anti-Swing Fuzzy Controller Applied in a 3D Crane System, ETASR
oscillation simulation results. Engineering, Technology & Applied Science Research, Vol. 2, No. 2,
pp. 196200, 2012.
[10] H.-H. Lee and S. Cho, A New Fuzzy-Logic Anti-Swing Control for
VI. CONCLUSION Industrial Three-Dimensional Overhead Cranes, 2001 IEEE
International Conference on Robotics & Automation, Seoul, Korea,
The results obtained from the mathematical model of the 2001, pp 2956 2961.
[11] M. B. Trabia, J. M. Renno and K. A. F. Noustafa, A General Anti-
three-dimensional crane show that the crane motion and the Swing Fuzzy Controller for an Overhead Crane with Hoisting, 2006
positioning can be tuned with the correction gain for every IEEE International Conference on Fuzzy Systems, Vancouver, BC,
axis. The state-space model behaves as expected and the axes Canada, 2006, pp. 627 634.
[12] Z. Nowacki, D. Owczarz and P. Wozniak, On the Robustness of Fuzzy
move according to the value of input force. As well, the study
Control of an Overhead Crane, Institute of Automation Control,
of the system errors shows that the angles of the payload still Technical University of Lodz, Poland, 1996, pp. 433437.
provide some mismatch in amplitude, though the oscillation [13] T. Popadic, F. Kolonic and A. Poljugan, A Fuzzy Control Scheme for
frequency for both and angels nevertheless is very similar. the Gantry Crane Position and Load Swing Control, Available at:
https://bib.irb.hr/datoteka/248540.MIPRO_FuzzySPG_V4.pdf
The system developed can be recommended for the accurate [Accessed: Oct. 18, 2015]
and safe 3D gantry crane control in industrial applications. [14] M. A. Ahmad, M. S. Ramil, R. M. T. Raja Ismail, A. N. K. Nasir and M.
Comparison of the anti-swing fuzzy logic controller for the A. Zawawi, The investigation of PD-type Fuzzy Logic with different
polarities input shaping for anti-sway control of a gantry crane system,
three-dimensional crane and the fuzzy logic five degree of 2009 Conference on Innovative Technologies in Intelligent System and
freedom controller show that the PI regulator is easier to Industrial Applications (CITISIA 2009), Sunway, Malaysia, 2009,
design comparing to fuzzy logic regulator. Though the pp. 452 457.
[15] C.-Y. Chang, Adaptive Fuzzy Controller of the Overhead Crane With
positioning in 5DOF fuzzy logic controller has an error, it is Nonlinear Disturbance, IEEE Transactions on Industrial Informatics,
still more stable and accurate comparing to PI controller. Vol. 3, No 2, pp. 164 172, 2007.
Furthermore, the anti-swing task is well carried by FLC [16] J. Smoczek and J. Szpytko, Fuzzy Logic-based Adaptive Control
comparing to anti-swing PI controller, what is in the current System Prototypying for Laboratory Scaled Overhead Crane, 18th
International Conference on Methods and Models in Automation and
case the goal of the research. Robotics (MMAR), Miedzyzdroje, Poland, 2013, pp 92 97.
The presented study opens the directions to the future [17] Y. Zhao and H. Gao, Fuzzy-Model- Based Control of an Overhead
research that may cover the finding of algorithms to scale the Crane With Input Delay and Actuator Saturation, IEEE Transaction on
Fuzzy Systems, Vol 20, No. 1, pp. 181 186, 2012.
fuzzy logic controller membership function universes of [18] S. Carmeli and M. Mauri, HIL test bench to test anti-swing fuzzy
discourse for several experimental objects. Different types of control of an overhead crane, 2013 IEEE International Conference on
cranes such as three-dimensional and tower cranes can be Mechatronics (ICM), Vicenza, Italy, 2013, pp. 754 760.
[19] C.-Y. Chang, K.-C. Hsu, K.-H. Chiang and G.-E. Huang, An Enhanced
explored using the offered algorithms. Self-scaling intelligent Adaptive Sliding Mode Fuzzy Control for positioning and Anti-Swing
fuzzy logic controller for the multidimensional motor driven Control of the Overhead Crane System, 2006 IEEE International
applications can be proposed in this way. Conference on System, Man, and Cybernetics, Taipei, Taiwan, 2006, pp.
992 997.
[20] L. Wang, H. Zhang and Z. Kong, Anti-swing Control of Overhead
REFERENCES Crane Based on Double Fuzzy Controller, 27th Chinese Control and
[1] R M. T. Raja Ismail, M. A. Ahmad, M. S. Ramli and F. R. M. Rashidi, Design Conference (CCDC), Qingdao, China, 2015, pp. 981 986.
Nonlinear Dynamic Modelling and Analysis of a 3-D Overhead Gantry [21] J. Goljevac, Comparison Between PID and Fuzzy Control, Ecole
Crane System with System Payload Variation, Third European Polytechnique Fdrale de Lausanne, Dpartement dInformatique,
Symposium on Computer Modeling and Simulation Athens, Greece, Laboratoire de Microinformatique, Internal Report R93.36I
2009, pp 350 354. [22] K. M. Passino and S. Yurkovich: Fuzzy Control. California: USA,
[2] H. Chen, B. Gao and X. Zhang, Dynamical Modelling and Nonlinear Addison Wesley Longman, Inc., 1998.
Control of a 3D Crane, 2005 International Conference on Control and
Automation (ICCA2005), Budapest, Hungary, 2005, pp. 1085 1090.

DOI: 10.1515/ecce-2015-0006
 13

Andrei Aksjonov received B.Sc and M.Sc in

electrical engineering and electrical drives and power
electronics with automation specialization from
Tallinn University of Technology (TUT) in 2013 and
2015, respectively. His current research topic is
Design and Implementation of Intellectual Fuzzy
Regulators and Observers for Multidimensional
Motor Driven Applications.

Postal address: Ehitajate Tee 5, 19086, Tallinn, Estonia.

E- mail: andrei.aksjonov@outlook.com

Valery Vodovozov received his Candidate o f

Science degree and Associate Professor and Senior
Researcher academic titles in Electrical Engineering
from St. Petersburg Electrotechnical University,
Russia. He is a professor of Tallinn University of
Technology, a member of IEEE, International
Institute of Informatics and Systemics (IIIS), and
Global Research Alliance of Texas Institute of
Science (TxIS GRA).

Postal address: Ehitajate Tee 5, 19086, Tallinn, Estonia.

E- mail: valery.vodovozov@ttu.ee

Eduard Petlenkov received his B.Sc, M.Sc and PhD

degrees in computer and systems engineering from
Tallinn University of Technology. He is an Associate
Professor in the Department of Computer Control at
Tallinn University of Technology. His main research
interests lie in the domain of nonlinear control,
system analysis and computational intelligence.

Postal address: Ehitajate Tee 5, 19086, Tallinn, Estonia.

E- mail: eduard.petlenkov@ttu.ee

DOI: 10.1515/ecce-2015-0006