Sei sulla pagina 1di 13

Mechanical Systems and Signal Processing 66-67 (2016) 756–768

Contents lists available at ScienceDirect

Mechanical Systems and Signal Processing


journal homepage: www.elsevier.com/locate/ymssp

An optimal performance control scheme for a 3D crane


Mohammad Javad Maghsoudi a, Z. Mohamed a,n, A.R. Husain a, M.O. Tokhi b
a
Faculty of Electrical Engineering, Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Johor, Malaysia
b
Department of Automatic Control and Systems Engineering, The University of Sheffield, Sheffield, United Kingdom

a r t i c l e in f o abstract

Article history: This paper presents an optimal performance control scheme for control of a three
Received 9 October 2014 dimensional (3D) crane system including a Zero Vibration shaper which considers two
Received in revised form control objectives concurrently. The control objectives are fast and accurate positioning of
26 March 2015
a trolley and minimum sway of a payload. A complete mathematical model of a lab-scaled
Accepted 25 May 2015
Available online 12 June 2015
3D crane is simulated in Simulink. With a specific cost function the proposed controller is
designed to cater both control objectives similar to a skilled operator. Simulation and
Keywords: experimental studies on a 3D crane show that the proposed controller has better
3D crane performance as compared to a sequentially tuned PID-PID anti swing controller. The
Input shaping
controller provides better position response with satisfactory payload sway in both rail
Optimal control
and trolley responses. Experiments with different payloads and cable lengths show that
PID controller
Sway reduction the proposed controller is robust to changes in payload with satisfactory responses.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Cranes are important and used in many industries, factories and ware houses to transfer heavy loads from one place to
another [1]. One of the significant factors affecting productivity and efficiency of the industrial systems is speed. However it
is obvious for a crane, fast manoeuvres resulted in significant payload oscillation and residual sway that negatively affects
performance of the system. At higher speeds, these sway angles prevent the payload to settle down during movement and
unloading. This problem will be crucial particularly for industrial applications where operators should manipulate the
cranes [2]. Vast applications of cranes have encouraged many researchers around the globe to reduce the motion-induced
oscillation of these structures [3]. One of the most useful approaches to reduce the motion-induced oscillations of oscillatory
systems is input shaping [4].
Initially, the input shaping theory was introduced in the late 1950s [5] and it was named as “Posicast Control”. The well-
known input shaping technique was firstly introduced by Singer and Seering [6]. Researchers have provided and utilised
different types of input shapers including Zero Vibration (ZV) and Zero Vibration Derivative (ZVD) [7], negative shapers [8],
multi-hump extra intensive shapers [9], two mode shapers [10] and unity magnitude shapers [11]. The first time input
shaping implemented on a real gantry crane was at Savanah Research Technology Centre [12] and researchers proved their
approach was successful to move the cart without sway during hoisting. Recently, implementing methods of input shaping
on cranes have been improved by researchers [13].

n
Corresponding author. Tel.: þ607 5557019; fax: þ607 5566272.
E-mail address: zahar@fke.utm.my (Z. Mohamed).

http://dx.doi.org/10.1016/j.ymssp.2015.05.020
0888-3270/& 2015 Elsevier Ltd. All rights reserved.
M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768 757

Despite the advent of many control theories and techniques, PID feedback controller is still one of the most widely used
control algorithm in industries [14]. They are easy to implement, low cost and effective and thus they have a significant role
in industrial applications. Many type of PID based control schemes to control the swing motion and cart position of the
cranes have been presented by researchers including neural network based PID controller [15], PSO based PID controllers
[16], genetic algorithm based PID controllers [17] and fuzzy based PID controllers [18–20]. Combination of input shaping and
PID controller has been previously proposed [21]. Some researchers also investigated this combination carefully [22] and a
closed-loop application of this combination has been also proposed [23]. Moreover, a comprehensive analysis on closed-
loop applications of input shaping has been done [24]. Recently, PID and S-shaped input are utilised to reduce the sway of a
rotary crane [25]. Some researchers discussed optimal closed loop PID–PD control of gantry crane without input shaping
[26]. However, an optimal performance closed-loop control scheme including both PID and input shaping has not been
proposed.
In this paper, a control scheme is presented for an automatic crane based on optimal design of PID and ZV shaper in a
closed loop system. A specific cost function to solve two control objectives concurrently is proposed. The objective is to
design a practical anti-swing controller without utilising the sway reduction loop of PID–PID control scheme. In the other
words, the PID controller gains of a closed-loop system are optimised based on a specific cost function which considers
actual position and sway of the payload. Thus a simple architecture with only one loop can satisfy the control requirements
in real time experiment. Initially, a complete mathematical model of the system is obtained based on Newtonian techniques.
The closed loop system including the ZV shaper, saturation block and PID controller is designed in Simulink. PID controller's
gains are then optimised to satisfy a specific cost function based on the absolute value of the difference between desired
position of payload and its actual position.
A PID–PID anti-swing controller [18] augmented with ZV shaper is considered for comparison. The result shows that the
proposed optimal performance controller performs better than the PID–PID anti-swing controller in several aspects.
Although, the proposed controller needs an accurate model for designing purpose and cannot handle environmental
disturbances, its simplicity and good performance may be useful. Based on this scheme, a crane can be moved to a
predefined destination with a reduced motion-induced sway of a payload.

2. Three dimensional (3D) crane system

Fig. 1 shows a lab-scaled 3D crane used in this study. The crane is capable of transferring a load from any location to a
desired place in a restricted three dimensional space. The system hardware consists of three main components: a cart, a rail
and a pendulum. The mathematical model is obtained based on the given characteristics of the crane by the manufacturer
and the study by Pauluk et al. [27]. The obtained model is simulated using Simulink to investigate dynamic behaviour of the
system.
A schematic diagram of the 3D crane system is shown in Fig. 2 with XYZ as the coordinate system. mp , mt and mr are the
payload mass, trolley mass (including gear box, encoders and DC motor) and moving rail respectively. l represents the length
of the lift-line, α represents the angle of lift-line with Y axis and β represents angle between negative part of Z axis and
projection of the payload cable onto the XZ plane. T is a reaction force in the payload cable acting on the trolley, Fx and Fy are
the forces driving the rail and trolley respectively, Fz is a force lifting the payload and fx, fy and fz are corresponding friction
forces. By defining

μ1 ¼ mmpt ; μ2 ¼ mtmþpmr ; Fx
u1 ¼ m t
F
; u2 ¼ mt þy mr ; u3 ¼ mF zp ; fx
f1 ¼m t
f
; f 2 ¼ mt þy mr ; f 3 ¼ mf zp

K 1 ¼ u1  f 1 ; K 2 ¼ u2  f 2 ; K 3 ¼ u3  f 3

Fig. 1. A lab-scaled 3D crane.


758 M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768

Fz
fy

fx
mt, trolley Fx

X
T
mr, rail l
α
Fy
fz β
-T

Y payload mp g

Fig. 2. Schematic diagram and forces.

the dynamic equations of motion of the crane can be obtained as [27]


x€ t ¼ K 2 þ μ2 K 3 sin α sin β ð1Þ

y€ t ¼ K 1 þ μ1 K 3 cos α ð2Þ

_ 2 lβ_ Þ sin α sin β þ 2lα


_ β_ cos α cos β
2
x€ p ¼ x€ t þ ð€l  lα

þ ð2_lα € Þ cos α sin β þ ð2_lβ_ þlβ€ Þ sin α cos β


_ þ lα ð3Þ

y€ p ¼ y€ t þð€l  lα
_ 2 Þ cos α ð2_lα
_ þlα
€ Þ sin α ð4Þ

_ 2 þlβ_ Þ sin α cos β þ 2lα


_ β_ cos α sin β
2
z€ p ¼ ð  €l þlα

 ð2_lα € Þ cos α cos β þ ð2_lβ_ þ lβ€ Þ sin α sin β


_ þ lα ð5Þ

where xp , yp and zp are position of payload in X, Y and Z axes respectively. xt and yt are positions of trolley in X and Y axes.
Dots represent derivative of the respective quantities. Table 1 shows the parameters used for simulation and experiment
which correspond to the lab-scaled crane shown in Fig. 1.

3. Particle swarm optimisation (PSO) and input shaping

In this paper, a proposed optimal controller is compared with PID–PID anti-swing control [18,26]. Optimal PID gains for
both controllers are obtained using PSO based on a certain cost function. To achieve higher sway reduction, ZV input shaper
is adopted in a closed-loop system for both controllers.

3.1. PSO

PSO is one of artificial intelligence families that was introduced in 1995 [28]. In PSO every particle finds a solution to the
optimisation problem. The position of a particle is changed according to the best position experienced by it and that of the
best particle in the entire swarm. The closeness of the particle to the global optimum is measured using the predefined
fitness function. Each particle in the group is defined in the following format:

 X i : the present position of the particle


 V i : the present velocity of the particle
 P i : the local best position of the particle

Based on the above definitions, the position of a particle changes according to:
V i ðk þ 1Þ ¼ iw U V i ðkÞ þr 1 Uac1 ðP i ðkÞ  X i ðkÞÞ þ r 2 U ac2 ðGi ðkÞ X i ðkÞÞ ð6Þ

X i ðk þ 1Þ ¼ X i ðkÞ þ V i ðk þ 1Þ ð7Þ
M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768 759

where r 1 and r 2 are positive random numbers produced by a uniform distribution and limited by an upper bound of 1.
Coefficients ac1 and ac2 are known as acceleration constants and iw is the inertia weight. Moreover, pi is the local best
solution found up to now by the i-th particle, while Gi shows the position of the best particle thus far in the entire swarm.
The local best position of particle i is the best position experienced by particle i up to now. If f is the fitness function then
the personal best position of a particle at iteration k can be written as
(
P i ðkÞ if f ðX i ðk þ1ÞÞ 4 f ðP i ðkÞÞ
P i ðk þ1Þ ¼ ð8Þ
X i ðk þ 1Þ if f ðX i ðk þ 1ÞÞ o f ðP i ðkÞÞ

After completing the algorithm, a large number of particles are predicted to converge within a small radius around the
global optimum of the flying space.
In this paper, PSO is used to tune the PID controllers of two different schemes including the proposed optimal
performance control and the PID–PID anti-swing control. For the first scheme, a special fitness function that considers both
position and sway concurrently is utilised. On the other hand for the PID–PID anti-swing control, initially PSO is used to tune
the PID controller for position control loop without sway consideration. Then with the obtained values for position control,
the other PID controller is tuned by minimising the motion induced sway of anti-swing loop. More descriptions on both
control schemes are given in Section 4. Table 2 shows PSO parameters used in this study.

3.2. Input shaping

Input shaping as shown in Fig. 3 is a feed-forward control technique that involves filtering a desired command with an
input shaper. A mathematical description of a general input shaper can be expressed as
X
m
ISðtÞ ¼ kj δðt t j Þ ð9Þ
j¼1

where δðtÞ represents the Dirac delta function, t j is time of jth pulse and a non-negative value and kj is amplitude of jth pulse
and a non-zero value. The shaped input that results from the convolution [6] will drive the system and the shaped command
reduces the detrimental effects of the oscillatory system. An oscillatory system can be modelled as a superposition of second
order systems each with a transfer function
ωn 2
GðsÞ ¼ ð10Þ
s2 þ 2ζωn s þ ωn 2
where ωn is the natural frequency and ζ is the damping ratio of the system.
To achieve zero vibration and to ensure that the shaped command input produces the same rigid body motion as the
unshaped command, a two-impulse sequence (Fig. 3) with parameters as
t 1 ¼ 0; t 2 ¼ ωπd ;
1 K
k1 ¼ ;
k2 ¼ ð11Þ
1þK 1þK
 pffiffiffiffiffiffiffiffiffi2ffi
is obtained where K ¼ e  ζπ = 1  ζ [5]. This is known as a ZV shaper.

Table 1
System parameters.

Variables Values

Mass of payload, mp 1 kg
Mass of trolley, mt 1.155 kg
Mass of moving rail, mr 2.2 kg
Cable length, l 0.47 m
Gravitational constant, g 9.8 m/s
Corresponding friction forces, f x ; f y ; f z 100, 82, 75 Ns/m

Table 2
PSO parameters for tuning controllers.

Variables Values

Integration time 10 s
Time interval 0.001 s
Population size 20
Inertia weight, iw 0.4–0.9
Number of iterations 130
Acceleration constant, ac1 ¼ ac2 1.42
760 M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768

Fig. 3. ZV input shaping.

4. Control schemes

The 3D crane system under consideration is a nonlinear system. Designing controllers for nonlinear systems normally
involves linearisation of the nonlinear equations. For this reason, in some operating points there would be considerable
differences between the performance of real and modelled systems. In this study, the controllers will be designed without
linearisation of its dynamical equation. For this purpose, an optimisation technique is utilised to tune the controller for the
3D crane system with nonlinear equations.
A control scheme which considers sway and position concurrently is proposed to have an optimal system performance.
This control scheme utilises only one control loop to cater both fast positioning and sway reduction. PID–PID anti-swing
control which utilises two control loops is then developed and used for performance comparison with the proposed
controller. Although both control schemes utilise PSO to find the gains of controllers, optimal performance control scheme is
simpler as compared to the PID–PID anti-swing controller. Moreover, the PID–PID controller needs an additional sensor to
measure the real time sway angle of the crane. The proposed scheme does not need any angular sensor in real time
operation and the optimisation process is performed only in simulation environment.

4.1. Optimal performance control scheme

In the proposed scheme an intelligent combination of PID controller, input shaping and a multi-objective optimisation
approach is proposed. Fig. 4 shows a block diagram of the optimal performance control scheme where Xdes and x are desired
and actual position responses respectively. e(t) is the system error and u(t) is the control input force. Based on the actual
displacement and angular displacement responses, PSO finds optimal PID gains which are proportional, K p , derivative, K d ,
and integral, K i , gains. The control input u(t) can be written as
 Z t 
de
uðtÞ ¼ K p eðtÞ þK d þ K i eðtÞdt ð12Þ
dt 0

The optimum solution can be obtained by using a proper fitness function. In this study, a new fitness function to minimise
sum of absolute error is considered as

X
Ts  
S¼ ½xj þ l  sin θj  X des  ð13Þ
i¼0

where T s is the time of simulation, j is j-th sample time of the simulation, xj is the absolute value of the position output in
the time interval of j, and θj is the absolute value of sway in the time interval of j. S is inspired by actual position of payload
and can be minimised by allocating different values for K p , K i and K d based on velocity and position formulas in PSO
described in Eqs. (6) and (7). Conceptually, PSO tries to find the optimum gains of PID controller to have the minimum error
signal similar to an experienced operator who can move the crane fast and without sway.
To simulate the shaper, a new design utilising delay units, gain units, flip–flop, intelligent switch and absolute value block
is proposed to produce the shaped command for every input automatically. Previously, gain–delay units were utilised to
shape a command for linear models [4,29]. However, as the design did not include a mechanism to detect rising and falling
edges, the approach was not able to produce a proper shaped command for falling edge of the input signal for a nonlinear
model as shown in Fig. 5. For this reason, the system produced undesired oscillation at the falling edge of the input signal.
In a closed-loop control scheme where control signal will be initially positive and suddenly changes to negative, the
design should also produce the shaped command in stopping time to prevent undesired oscillation at final position. As
shown in Fig. 6, the proposed system can automatically detect the rising and falling edges of the input regardless of its
direction and sets the latch flip–flop on and off respectively. This causes the intelligent switch to select k1 and t 1 using Eq.
(11) for rising and falling edges of the input respectively. Although, the proposed scheme is simulated in Simulink, it is
feasible to be implemented simply in an industrial programmable logic controller.
M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768 761

Fig. 4. PSO-tuned PID control scheme.

1
Unshaped Input
0.8 Gain Delayed Shaped
Proposed Shaped
0.6

0.4

0.2
Force (N)

-0.2

-0.4

-0.6

-0.8

-1
0 2 4 6 8 10 12 14 16 0 2
Time (s)

Fig. 5. Unshaped, gain delayed shaped and proposed shaped signals.

-K -

k1

-K -

t1 (1-k1)

SQ
Edge Detector
1 |u| R!Q 1
Control Signal Shaped Signal
Abs S-R Switch
Flip -Flop
Edge Detector 1

-K -

(1-k1).

-K -

t1. k1.

Fig. 6. Proposed ZV shaper.

4.2. PID–PID anti-swing control

The main objectives of PID–PID anti-swing control are to move the payload from a specific point to a desired destination
fast and without considerable swing at the final position. Researchers commonly use two controllers [18,23,26,30,31] to
control both trolley position and swing angle of the payload of a gantry crane as shown in Fig. 7 where Kps, Kis and Kds are
PID parameters for the inner loop. An outer loop PID controller is used for trolley positioning control and inner loop PID
controller is utilised for swing control of the payload. In this paper, ZV shaper is also included in the closed-loop system to
improve the system performance. To design the controllers, initially the PID gains of inner loops are set to zero and the outer
loop's PID controller including ZV shaper is designed to have the best positioning performance without sway consideration.
762 M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768

Fig. 7. PSO-tuned PID–PID control scheme.

1
PID-PID
0.8 PID Unshaped
PID Shaped
0.6

0.4
Force (N)

0.2

-0.2

-0.4

-0.6

-0.8

-1
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 8. Control signals of rail (simulation).

Table 3
Controller gains tuned using PSO.

System PID–PID Optimal PID

Kp Ki Kd K ps K is K ds Kp Ki Kd

Rail 35.9 0.0007  0.13 3.75 6.9  0.584 3.9042 0.1419 0.3828
Trolley 35.55 0.1162  0.96 0.99 5.428 0.1653 3.7454 0.1556 0.3399

Then, the obtained gains of outer loop are fixed and the PID controller of the inner loop is designed sequentially to minimise
the motion induced sway of payload. Both controllers are designed based on PSO.

5. Simulation results

In this section, the control schemes are implemented and tested within the simulation environment of the 3D crane
system and the corresponding results are presented. In this simulation, a personal computer with Intel Core2 Duo CPU and
2.1 GHz clock frequency was utilised. The Simulink is utilised to implement the optimal performance control scheme.
Initially, all the particles and gains of controller are randomly initialised. The simulation time is set to 10 s and the fitness
function is computed at each iteration until a predefined value of iteration is reached. The PSO parameters and gains for
both controllers for rail and trolley are shown in Tables 2 and 3 respectively. The PID–PID controller and optimal
performance PID controller are simulated based on the obtained gains to move the crane to the desired position, Xdes.
Figs. 8 and 9 show simulation results of the rail and trolley control signals, respectively. It can be shown that the optimal
performance PID control scheme needs less effort of actuator compared to PID–PID control scheme. Moreover, the proposed
ZV shaper design has shaped the control signal of PID controller in both falling and rising edges properly. The position
responses of both control schemes for rail and trolley are shown in Figs. 10 and 11 respectively. Considering the input as
desired response and zero as desired sway, Integrated Absolute Error (IAE) values for rail and trolley are calculated for both
controllers. Table 4 summarises the settling time, IAE values and overshoot of position responses using both controllers for
rail and trolley. It can be shown that for both rail and trolley, the proposed control scheme provides better position response
M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768 763

1
PID-PID
0.8 PID Unshaped
PID Shaped
0.6
0.4

Force (N)
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 9. Control signals of trolley (simulation).

0.5
0.45
0.4
0.35
Position (m)

0.3
0.25
0.2
0.15
0.1 Input
PID-PID
0.05
Optimal PID
0
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 10. Position of rail (simulation).

0.5
0.45
0.4
0.35
Position (m)

0.3
0.25
0.2
0.15
0.1 Input
0.05 PID-PID
Optimal PID
0
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 11. Position of trolley (simulation).

Table 4
Settling time, overshoot and IAE values of system responses.

System Settling time (s) IAE Overshoot (%) IAE (Sway)

Optimal PID PID–PID Optimal PID PID–PID Optimal PID PID–PID Optimal PID PID–PID
Rail 1.76 4.48 36.03 45.14 0 13.68 1625 1289
Trolley 2.3 4.66 43.64 48.39 0 6.13 454 754

performance as compared to PID–PID controller. With the proposed controller, the position response show no overshoot in
both directions of the crane, lower IAE values (20% for rail and 10% for trolley) and faster settling time which is nearly half of
PID–PID controller.
764 M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768

8
Optimal PID
6 PID-PID

Sway (deg)
2

-2

-4

-6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 12. Sway of payload in X direction (rail).

6
Optimal PID
PID-PID
4

2
Sway (deg)

-2

-4

-6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 13. Sway of payload in Y direction (trolley).

1
PID-PID
0.8 PID Unshaped
0.6 PID Shaped

0.4
0.2
Force (N)

0
-0.2
-0.4
-0.6
-0.8
-1
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 14. Control signal of rail (experiment).

Figs. 12 and 13 show the sway response of rail and trolley for both controllers. The results show that optimal PID
controller provides better transient sway for both rail and trolley. With the optimal PID, maximum transient sways are 51
and 41 respectively whereas maximum transient sways of 7.51 and 5.51 using PID–PID controller are noted. However, it is
noted that the PID controller yields higher residual sway for rail as compared to PID–PID controller. Nevertheless, for such a
very low damped system where ζ is determined by experiments as 0.008, the residual sway of 21 is very small near to zero.
For trolley's payload sway, almost similar residual sway is achieved with both controllers. Generally, it can be stated that
considering both objectives concurrently to have an optimum response with only one PID controller can provide satisfactory
results compared to the PID–PID control scheme which utilises two PID controllers in two loops. The results are reflected
with IAE values shown in Table 4 where the proposed controller provides 30% improvement for trolley.
M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768 765

1
PID-PID
0.8 PID Unshaped
0.6 PID Shaped

0.4

Force (N)
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 15. Control signal of trolley (experiment).

0.5

0.4
Position (m)

0.3

0.2

0.1 Input
Optimal PID
PID-PID
0
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 16. Position of rail (experiment).

6. Experimental results

Experiments were performed on a lab-scaled 3D crane shown in Fig. 1 to verify the simulation results. The crane is
equipped with 5 incremental encoders to measure the position of trolley in X and Y directions, position of payload in Z
direction and the payload sway in X and Y directions. All the data are transmitted on-line to a PC by an interface card. The
motors are also actuated by specific DC drives. Similar parameters used in simulation including input shaping parameters
and controller gains are applied to the real 3D crane. Figs. 14 and 15 show control signals for the rail and trolley respectively.
The results verify the simulation results where the proposed optimal control needs less actuator effort as compared to PID–
PID controller. Moreover, the control signal is smoother.
Experimental position responses of both control schemes for rail and trolley are shown in Figs. 16 and 17 respectively.
Table 5 shows the settling time, IAE values and overshoot of the position responses for rail and trolley. The results also verify
the simulation results where the optimal PID gives better performance in the position response. There is no overshoot for
PID controllers in both X and Y directions, IAE values are 25% and 14% lower for rail and trolley respectively, and their
settling times are nearly half of PID–PID controller.
The sway outputs for both X and Y directions which are captured by two angle incremental encoders of lab-scaled crane
are shown in Figs. 18 and 19. Similar results as the simulation are obtained where the optimal PID controller provide better
transient response sway with maximum transient sways as 4.51 and 4.21 for rail and trolley respectively. With the PID–PID
controller, maximum transient sways are 51 and 6.11 respectively. It is found that the optimal PID controller provides slightly
higher residual sway of rail that can also be neglected. However, less residual sway of trolley response is shown with the
controller. Nevertheless, the IAE values for sway are almost 30% lower with the optimal controller for both responses.

7. Robustness

For evaluation of robustness, the 3D crane with different payloads and cable lengths are considered. The previous case
with the payload, mp ¼1 kg, cable length, l¼0.47 m and sway frequency of 4.57 rad/s is used as a reference. By using similar
PID gains and ZV shaper parameters obtained with the reference case, robustness of the proposed controller to changes in
payload and cable length of the crane can be examined. For changes in payloads, experiments were conducted with 0.4 kg
766 M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768

0.5
0.45
0.4
0.35

Position (m)
0.3
0.25
0.2
0.15
0.1 Input
Optimal PID
0.05
PID-PID
0
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 17. Position of trolley (experiment).

6
Optimal PID
PID-PID
4

2
Sway (deg)

-2

-4

-6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 18. Experimental sway of payload in X direction (rail).

8
Optimal PID
6 PID-PID

4
Sway (deg)

-2

-4

-6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 19. Experimental sway of payload in Y direction (trolley).

Table 5
Settling time, overshoot and IAE values of system responses (experiment).

System Settling time (s) IAE Overshoot (%) IAE (Sway)

Optimal PID PID–PID Optimal PID PID–PID Optimal PID PID–PID Optimal PID PID–PID
Rail 2.17 6.02 41.73 56.09 0 7.12 700 1006
Trolley 2.15 4.76 43.01 50.43 0 7.27 724 1153
M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768 767

0.45
0.4
0.35

Position (m)
0.3
0.25
0.2
0.15
l=0.326 m, mp=1 kg
0.1 l=0.730 m, mp=1 kg
0.05 l=0.47 m, mp=1.3 kg
l=0.47 m, mp=0.73 kg
0
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 20. Position of trolley with changes in natural frequency and weight.

8
l=0.326 m, mp=1 kg
6 l=0.73 m, mp=1 kg
l=0.47 m, mp=1.3 kg
4 l=0.47 m, mp=0.73 kg
Sway (deg)

-2

-4

-6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Fig. 21. Sway of payload with changes in natural frequency and weight (trolley).

and 1.3 kg that implies  60% and þ30% of the reference payload. Subsequently, cable lengths of 0.326 m and 0.73 m were
investigated that correspond to payload sway frequencies of 5.49 rad/s and 3.66 rad/s respectively. These imply 720%
difference from the reference frequency.
Fig. 20 shows trolley position responses for all cases subjected to a desired input position of 0.4 m as in the previous
experiments. The results show that, using different payloads and cable lengths, uniform performance of the trolley position
as the reference case (Fig. 17) was achieved. Fig. 21 shows experimental sway response of the crane with different payloads
and cable lengths. By comparing with the reference case in Fig. 18, it is noted that the proposed controller can handle
different payloads as sway response are not affected. However changing the cable length and sway frequency affects
significantly the sway response. The result is justified as a non-robust ZV is used [32] and thus the robustness could be
increased with a robust shaper.

8. Conclusion

An optimal performance control scheme has been developed to enable a 3D crane to reach to a predefined target faster,
without overshoot and less motion-induced oscillation. The control scheme utilises a non-analytical approach to find
optimal controller's gains. An automatic input recognising scheme which is feasible to be implemented by industrial
programmable logic controllers is also presented to shape the input signal in both falling and rising edges without using
inner-loop PID controller for payload sway reduction. Simulation and experimental results verify that the controller is able
to improve position responses with acceptable payload sway of rail and trolley of a 3D crane as compared to PID–PID anti
swing controller. The proposed control scheme has also been shown to be robust to different payloads and cable lengths of
the crane.

References

[1] H. Butler, G. Honderd, J. Van Amerongen, Model reference adaptive control of a gantry crane scale model, IEEE Control Syst. 11 (1991) 57–62.
[2] K.C.C. Peng, W. Singhose, D.H. Frakes, Hand-motion crane control using radio-frequency real-time location systems, IEEE/ASME Trans. Mechatron. 17
(2012) 464–471.
768 M.J. Maghsoudi et al. / Mechanical Systems and Signal Processing 66-67 (2016) 756–768

[3] J. Smoczek, Fuzzy crane control with sensorless payload deflection feedback for vibration reduction, Mech. Syst. Signal Process. 46 (2014) 70–81.
[4] W. Singhose, Command shaping for flexible systems: a review of the first 50 years, Int. J. Precis. Eng. Manuf. 10 (2009) 153–168.
[5] O.J.M. Smith, Posicast control of damped oscillatory systems, Proc. IRE 45 (1957) 1249–1255.
[6] N. Singer, W. Seering, Preshaping command inputs to reduce system vibration, ASME J. Dyn. Syst. Meas. Control 112 (1990) 76–82.
[7] W.E. Singhose, W.P. Seering, N.C. Singer, Shaping inputs to reduce vibration: a vector diagram approach, in: Proceedings of the IEEE International
Conference on Robotics and Automation, Cincinnati, OH, 1990, pp. 922–927.
[8] W. Singhose, N. Singer, W. Seering, Design and implementation of time-optimal negative input shapers, in: ASME Winter Annual Meeting, Chicago, IL,
1994, pp. 151–157.
[9] W. Singhose, S. Derezinski, N. Singer, Extra-insensitive input shapers for controlling flexible spacecraft, J. Guid. Control. Dyn. 19 (1996) 385–391.
[10] E.A. Crain, W.E. Singhose, W.P. Seering, Derivation and properties of convolved and simultaneous two-mode input shapers, IFAC World Congress, San
Francisco, CA, 1996, pp. 441–446.
[11] L. Pao, W. Singhose, Unity magnitude input shapers and their relation to time-optimal control, in: IFAC World Congress, San Francisco, CA, 1996,
pp. 385–390.
[12] N. Singer, W. Singhose, E. Kriikku, An input shaping controller enabling cranes to move without sway, in: ANS 7th Topical Meeting on Robotics and
Remote Systems, Westinghouse Savannah River Co., Aiken, SC, United States, 1997, pp. 225–231.
[13] M.J.C. Ronde, M.G.E. Schneiders, E.J.G.J. Kikken, M.J.G. van deMolengraft, M. Steinbuch, Model-based spatial feedforward for over-actuated motion
systems, Mechatronics 24 (2014) 307–317.
[14] G. Ellis, Control system design guide: using your computer to understand and diagnose feedback controllers, Butterworth-Heinemann, Oxford, UK,
2012.
[15] H. Saeidi, M. Naraghi, A.A. Raie, A neural network self tuner based on input shapers behavior for anti sway system of gantry cranes, J. Vib. Control 19
(2013) 1936–1949.
[16] M.I. Solihin, M. Kamal, A. Legowo, Optimal PID controller tuning of automatic gantry crane using PSO algorithm, in: Proceedings of the IEEE
International Symposium on Mechatronics and Its Applications, Amman, Jordan, 2008, pp. 1–5.
[17] M.I. Solihin, M. Kamal, A. Legowo, Objective function selection of GA-based PID control optimisation for automatic gantry crane, in: Proceedings of the
IEEE International Conference on Computer and Communication Engineering, Kuala Lumpur, Malaysia, 2008, pp. 883–887.
[18] M.I. Solihin, A. Legowo, P.I.D. Fuzzy-tuned, anti-swing control of automatic gantry crane, J. Vib. Control 16 (2010) 127–145.
[19] X. Li, W. Yu, Anti-swing control for an overhead crane with fuzzy compensation, Intell. Autom. Soft Comput. 18 (2012) 1–11.
[20] C.-N. Ko, A. Fuzzy PID controller based on hybrid optimisation approach for an overhead crane, in: Next Wave in Robotics: Proceedings of 14th FIRA
RoboWorld Congress, Kaohsiung, Taiwan, 2011, pp. 202–205.
[21] Z. Mohamed, M.A. Ahmad, Hybrid input shaping and feedback control schemes of a flexible robot manipulator, World Congress, Seoul, 2008,
pp. 11714–11719.
[22] J.R. Huey, The Intelligent Combination of Input Shaping and PID Feedback Control (Ph.D. thesis), Georgia Institute of Technology, Atlanta, 2006.
[23] K.L. Sorensen, W. Singhose, S. Dickerson, A controller enabling precise positioning and sway reduction in bridge and gantry cranes, Control Eng. Pract.
15 (2007) 825–837.
[24] J.R. Huey, K.L. Sorensen, W.E. Singhose, Useful applications of closed-loop signal shaping controllers, Control Eng. Pract. 16 (2008) 836–846.
[25] N. Uchiyama, H. Ouyang, S. Sano, Simple rotary crane dynamics modeling and open-loop control for residual load sway suppression by only horizontal
boom motion, Mechatronics 23 (2013) 1223–1236.
[26] H.I. Jaafar, Z. Mohamed, A.F.Z. Abidin, Z.A. Ghani, PSO-tuned PID controller for a nonlinear gantry crane system, in: Proceedings of the IEEE
International Conference on Control System, Computing and Engineering Penang, Malaysia, 2012, pp. 515–519.
[27] M. Pauluk, A. Korytowski, A. Turnau, M. Szymkat, Time optimal control of 3D crane, in: Proceedings of the 7th IEEE International Conference on
Methods and Models in Automation and Robotics, Międzyzdroje, 2001, pp. 122–128.
[28] J. Kennedy, R. Eberhart, Particle swarm optimisation, in: Proceedings of IEEE International Conference on Neural Networks, Perth, Australia, 1995,
pp. 1942–1948.
[29] M. Alam, M. Tokhi, Design of command shaper using gain–delay units and particle swarm optimisation algorithm for vibration control of flexible
systems, in: Proceedings of the 14th International Congress on Sound and Vibration, Australia, 2007, pp. 9–11.
[30] E.M. Abdel-Rahman, A.H. Nayfeh, Z.N. Masoud, Dynamics and control of cranes: a review, J. Vib. Control 9 (2003) 863–908.
[31] A. Ridout, Anti-swing control of the overhead crane using linear feedback, J. Electr. Electron. Eng. 9 (1989) 17–26.
[32] J. Vaughan, A. Yano, W. Singhose, Comparison of robust input shapers, J. Sound Vib. 315 (2008) 797–815.