Mechanical Systems and Signal Processing 46 (2014) 70–81

Contents lists available at ScienceDirect

Mechanical Systems and Signal Processing

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

Fuzzy crane control with sensorless payload deflection

feedback for vibration reduction
Jaroslaw Smoczek n
AGH University of Science and Technology, Faculty of Mechanical Engineering and Robotics, 30 Mickiewicza Av., 30-059 Krakow, Poland

a r t i c l e in f o abstract

Article history: Different types of cranes are widely used for shifting cargoes in building sites, shipping
Received 14 August 2013 yards, container terminals and many manufacturing segments where the problem of fast
Received in revised form and precise transferring a payload suspended on the ropes with oscillations reduction is
30 October 2013
frequently important to enhance the productivity, efficiency and safety. The paper
Accepted 23 December 2013
presents the fuzzy logic-based robust feedback anti-sway control system which can be
Available online 18 January 2014
applicable either with or without a sensor of sway angle of a payload. The discrete-time
Keywords: control approach is based on the fuzzy interpolation of the controllers and crane dynamic
Sensorless anti-sway control model0 s parameters with respect to the varying rope length and mass of a payload.
Fuzzy logic
The iterative procedure combining a pole placement method and interval analysis of
Robust control
closed-loop characteristic polynomial coefficients is proposed to design the robust control
Interval analysis
scheme. The sensorless anti-sway control application developed with using PAC system
with RX3i controller was verified on the laboratory scaled overhead crane.
& 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Reduction of the vibration is a serious concern for many flexible systems ranging from flexible manipulators arms [1,2] to
large industrial cranes which are extensively used for shifting goods in building sites, shipping yards, container terminals
and many manufacturing segments [3,4]. Different types of cranes (e.g. container cranes, overhead cranes, tower cranes, jib
cranes) realize the transportation operations through moving the cargos suspended at the end of flexible ropes. Thus, the
problem of positioning accuracy, effectiveness and safety of cranes operations requires to implement a control system for
reducing the pendulation effect.
Numerous researchers have proposed either the open-loop or closed-loop control solutions for vibration reduction,
frequently addressing them to an anti-sway crane control problem. The feedback control strategy requires installing a
reliable measurement system, that is sometimes difficult to maintain and costly. On the other hand, the open-loop anti-
sway crane control systems, developed usually based on the input shaping [5–7] or optimal control theory [8,9], are
sensitive to the disturbances owing to the lack of sway angle of a payload feedback. Hence, some works combine the
open-loop control scheme with the feedback solutions to reduce swing induced by both the human operator and external
disturbances [10].
The soft computing techniques, especially fuzzy logic, are widely employed in the closed-loop control schemes of flexible
dynamic systems. The linguistic-rule-based fuzzy controllers are reported in [11–13]. Fuzzy logic-based strategies are
proposed for PID gains tuning [14,15] and sliding mode control [16]. Takagi-Sugeno-Kang-type fuzzy controllers are

n
Tel.: þ48 126173104.
E-mail address: smoczek@agh.edu.pl

0888-3270/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ymssp.2013.12.012
 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81 71

proposed in Refs. [17–19]. Some researchers have adopted off-line or on-line techniques to design the fuzzy rule-based
controller implemented in a crane control scheme. Membership functions tuning techniques have been developed based on
an inverse dynamic [20], gradient algorithm [21], genetic algorithm [22,23], fuzzy clustering methods [19], and through
applying artificial neural network [18,24].
The most of the fuzzy logic-based approaches to an anti-sway crane control problem, which are described in the
literature are the linguistic rule-based strategies. The proposed techniques of fuzzy controller designing are mostly applied
for only tuning the membership functions parameters for a fixed number of rules involving the training data examples. The
robustness of crane control system is also frequently analyzed taking into account only the rope length variation.
Furthermore, most of the control strategies are only proved using mathematical models or mechatronic laboratory models.
Thus, there is still place for those researchers who are looking for the efficient control laws, software-hardware solutions
and measurement equipment of crane control systems that could be implemented in the industrial practice.
This paper proposes a fuzzy logic-based robust feedback anti-sway control system which can be applicable either with or
without a sensor of sway angle of a payload. The novelty approach to closed-loop anti-sway crane control with sensorless
payload deflection feedback, as well as the interval analysis-based design of fuzzy logic-based control scheme is presented.
The sensorless anti-sway discrete-time control approach is based on the feedback signal of payload deflection estimated by
a pendulum model formulated as the discrete-time relation between the sway angle of a payload and crane speed. The fuzzy
system is applied to interpolate the linear controllers and crane dynamic model parameters with respect to the varying rope
length and mass of a payload. The iterative procedure combining a pole placement method and interval analysis of closed-
loop characteristic polynomial coefficients was developed to design a fuzzy robust control scheme. The objective of a robust
control system design is to find minimum number of operating points at which the discrete-time dynamic crane model
parameters are estimated through the open-loop identification experiments and the linear controllers parameters are
determined using a pole placement method. The method of designing a fuzzy logic-based scheduling system leads to
determine the suitable number of fuzzy rules and fuzzy sets on the scheduling variables universe of discourse and
distribution of triangular-shaped membership functions parameters, which satisfy the acceptable range of performances
deterioration within the expected range of system parameters variation. The proposed method was applied to design and
implement the sensorless anti-sway control system on the laboratory scaled overhead crane using PAC system with RX3i
controller.
The paper is organized as follows. Section two describes a fuzzy logic-based discrete-time closed-loop control scheme for
a planar model of a crane. In section three, the iterative procedure used to design a complete and coherent rules base (RB) of
a fuzzy scheduler is proposed. Section four presents the experimental results obtained on the laboratory scaled overhead
crane. Section five delivers the final conclusions.

2. Fuzzy interpolation-based control scheme

The proposed fuzzy logic-based adaptive control scheme of 2D crane is presented in Fig. 2. The sensorless anti-sway
crane control system is based on the feedback signals of crane position and speed, rope length, mass of a payload, and sway
angle of a payload estimated by a pendulum model assumed as the second-order discrete-time transmittance representing

Fig. 1. Control scheme of 2D crane with estimation of sway angle of a payload and fuzzy logic-based interpolation of controllers and pendulum model
^ sway angle of a payload, and its
parameters, where: m – mass of a payload, l – rope length, xr, x – desired and current crane position, respectively, α; α-
estimated value, respectively, u – controlling signal.
72 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81

Fig. 2. Membership functions defined for fuzzy sets on the l and m input variables.

the relation between the payload angular deflection α and crane speed Vx. The controllers and model parameters are
interpolated by a fuzzy system based on the scheduling variable: rope length l and mass of a payload m.
The fuzzy interpolation scheme is composed of a set of N rules (1) with the singleton type conclusions representing the
controllers and model parameters determined at operating point associating with the centre points of triangular-shaped
membership functions (Fig. 2) used to evaluate the membership degree of l and m crisp input values to fuzzy sets specified
in the antecedent of a rule:
" #
Kk
Rk : If l is Li and m is Mj Then yk is ; ð1Þ
Θ2k

where yk is the vector of rule output (where k¼1, 2,…, N) and Kk and Θ2k are, respectively, the vectors of controllers and
pendulum model parameters determined at operating point (li, mj) associating with the centre points of triangular
membership functions of fuzzy sets Li and Mj (Fig. 2) on l and m input variables universe of discourse, where i¼1, 2,…, n and
j¼ 1, 2,…, r (n and r are the numbers of fuzzy sets defined for l and m, respectively).
The fuzzy sets Li and Mj correspond to the triangular membership functions (2) and (3) with the centre points [l1, l2,…, ln]
and [m1, m2,…, mr] distributed within the expected range of scheduling variables changes.
   
l li  1 li þ 1  l
μLi ðlÞ ¼ max min ; ;0 ; ð2Þ
li  li  1 li þ 1  li
where: li  1 rli r li þ 1 ; i ¼ 1; 2; n:
   
m  mj  1 mj þ 1  m
μMj ðlÞ ¼ max min ; ;0 ; ð3Þ
mj  mj  1 mj þ 1  mj

where: mj  1 r mj r mj þ 1 ; j ¼ 1; 2; ……; r:
The output vector of a fuzzy scheduler is calculated as the weighted average of all rules output:
" # " #! !1
K N Kk N
y¼ ¼ ∑ wk ∑ wk ; ð4Þ
Θ2 k¼1 Θ2k k¼1

where a rule0 s activation degree (weight of a rule) is calculated as follows:

wk ¼ μLi ðlÞμMj ðmÞ ð5Þ

In the published works, the crane systems are mostly modeled based on a lumped-mass [12,25] or distributed-mass
[26,27] approach. The most widely used lumped-mass models are based on the assumption that a payload is a point-mass
suspended at the end of a massless rigid rod. Abdel-Rahman et al. in a crane review article [28] classify the lumped-mass
models into reduced and extended models, in which the moving load influence on the dynamic response of cranes is either
taken into consideration or not, respectively. The most of the anti-sway control schemes reported in the literature are
developed based on the crane dynamic models derived from the Euler-Lagrange equation. In this paper, the crane dynamic
system is modeled as a lumped-mass planar model of a crane which is simplified to the first and second-order discrete-time
transfer functions, which describe the relation between crane speed and input function (6), and sway angle of a payload and
 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81 73

Fig. 3. Discrete-time closed-loop control system.

crane speed (7), where model0 s parameters vary in relation to the rope length l and mass of a payload m.
V x ðz; l; mÞ d0 ðl; mÞ
G1 ðz; l; mÞ ¼ ¼ : ð6Þ
Uðz; l; mÞ z þ c0 ðl; mÞ

αðz; l; mÞ b1 ðl; mÞz þb0 ðl; mÞ

G2 ðz; l; mÞ ¼ ¼ 2 : ð7Þ
V x ðz; l; mÞ z þa1 ðl; mÞz þa0 ðl; mÞ

The closed-loop control system (Fig. 3) was based on the linear controllers for crane position, speed and first-order
discrete-time controller of payload sway angle with parameters KTk ¼[k1, k2, q0, q1, s0]k determined for a set of N operating
points associated with the centre points (li, mj) of membership functions (Fig. 2).
Considering the above assumptions, the closed-loop control system design involves to select the suitable number of
operating points corresponding to the midpoints (li, mj) of triangular-shaped membership functions at which the
 T
parameters of a crane dynamic model Θ1k ¼ c0 ; d0 k and Θ2k ¼ ½a1 ; a0 ; b1 ; b0 Tk presented in the form of discrete-time
transmittances (6) and (7) are identified in the open-loop experiments, and the controllers parameters
 T
Kk ¼ k1 ; k2 ; q0 ; q1 ; s0 k are determined. Thus, the objective of closed-loop control synthesis is to find the number of fuzzy
sets and distribution of the membership functions parameters ensuring the desired performances in the expected range of
parameters variation.

3. The iterative procedure of designing the fuzzy interpolation scheme

The closed-loop control system (Fig. 3) transfer function can be presented in the following form:
0 1
ðb1 ðl; mÞz2
B þðb ðl; mÞs þ b ðl; mÞÞz C
k1 ðl; mÞk2 ðl; mÞd0 ðl; mÞ@ 1 0 0 A
αðz; l; mÞ þb0 ðl; mÞs0 Þ
¼ ; ð8Þ
X r ðzÞ z5 þ zSðl; mÞRðl; mÞ

where z¼[z4, z3, z2, z1, z0], S is an eliminant matrix consisting of model parameters and R is a vector containing of the
controllers parameters interpolated by a fuzzy system.
The RB of a fuzzy scheduler can be determined through assigning the closed-loop control system poles at the midpoints
of desired poles intervals. The objective of a fuzzy scheduler is to place all the closed-loop control system characteristic
polynomial coefficients within desired intervals. Thus, a set of operating points associating with the centre points of
membership functions, which satisfy the robust performances, can be obtained based on the objective function derived
from the interval Diophantine equation. Considering that all poles zf (where f¼1, 2,…, 5) of a closed-loop control system at
each k ¼1, 2,…, N operating point (li, mj) are assigned at the midpoints of the real numbers intervals representing a desired
region of stable poles denoted as:
½zf k ¼ ½zf ; zfþ k ¼ fzf A ℜjzf rzf r zfþ g; ð9Þ

the performances of a fuzzy logic-based control system satisfy desired conditions if the coefficients of closed-loop system
characteristic equation lie within the coefficients intervals of desired polynomial (10) determined using the arithmetic
operations on intervals [29]:
5
P k ðzÞ ¼ ∏ ðz  ½zf ; zfþ k Þ ¼ z5 þ z½P k ; ð10Þ
f ¼1

where [Pk] is an interval vector of desired characteristic equation coefficients:

½P k  ¼ ½½p4 k ; ½p3 k ; ½p2 k ; ½p1 k ; ½p0 k T : ð11Þ

74 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81

Hence, the controllers parameters vector Kk defined in a rule conclusion (1) can be determined based on the equations
system:
Sk R k ¼ P k ; ð12Þ

where:
2 ! ! 3T
a0  a1 þ c0 ða0 a1 Þ
6 a1 þc0 1 a0 c0 0 7
6 c0 ða1 1Þ  a0 7
6 7
6 7
6 d0 d0 ða1  1Þ d0 ða0  a1 Þ  d0 a0 0 7
6 7
6 0 0  d0 b1 d0 ðb1  b0 Þ  d0 b0 7
6 7
6 7
6 0  d0 b1 d0 ðb1  b0 Þ d0 b0 0 7
Sk ¼ 6
6
! ! 7
7
6 a0  a1 þ c0 ða0  a1 Þ 7
6 1 a1 þ c0  1 a0 c0 7
6 c0 ða1  1Þ a0 7
6 7
6 7
6 T s d0 T s d0 a1 T s d0 a0 0 0 7
6 7
6 0 d0 d0 ða1  1Þ d0 ða0  a1 Þ d0 a0 7
4 5
0 T s d0 T s d0 a1 T s d0 a0 0

Rk ¼ ½1; k2 ; q0 ; q1 ; s0 ; k1 k2 ; k2 s0 ; k1 k2 s0 T ; and Pk consists of the interval vector [Pk] midpoints.

The robust performances objective function derived from the equations system (12) is defined as:
Sðl; mÞRðl; mÞ A ½P k : ð13Þ

Therefore, the fuzzy logic-based control scheme satisfies the desired performances for the system parameters varying
 þ
within the expected ranges of rope length l A ½l ; l  and mass of a payload m A ½m  ; m þ , if the condition (13) is not
violated for the interval vectors (11) associating with the rules which has been activated to interpolate the controller
parameters with the firing strength factor wk 40.
The iterative procedure, described in this section, allows to determine minimum number of fuzzy sets on l and m
scheduling variables universe of discourse, and determine the complete and coherent RB of a fuzzy interpolator. The
proposed algorithm involves to identify the parametric models (6) and (7) of crane dynamic at operating points
corresponding to the lower and upper bounds of scheduling variables intervals [l  , l þ ] and [m  , m þ ], and to assume the
centre points of triangular-shaped membership functions as l1 ¼ l  , ln ¼l þ , m1 ¼m  and mr ¼m þ . Assuming the desired
poles intervals (9) for each operating point (l1, m1), (l1, mr), (ln, m1) and (ln, mr), the vectors of controller parameters are
derived from (12), and the initial set of rules is formulated as follows:

R1 : If l is L1 and m is M1 Then y1 is ½KT1 ; ΘT11 ; ΘT21 T

R2 : If l is L1 and m is Mr Then y 2 is ½KT2 ; ΘT12 ; ΘT22 T

R3 : If l is Ln and m is M1 Then y3 is ½KT3 ; ΘT13 ; ΘT23 T
R4 : If l is Ln and m is Mr Then y4 is ½KT4 ; ΘT14 ; ΘT24 T ð14Þ

The output vector of a fuzzy interpolating system is calculated as the weighted average of all rules output:
2 3 0 2 31
K Kk !1
N N
6Θ 7 B 6 Θ 7C
y ¼ 4 1 5 ¼ @ ∑ wk 4 1k 5A U ∑ wk ; ð15Þ
k¼1 k¼1
Θ2 Θ2k

The iterative design of a fuzzy interpolation control scheme is a two-stage process in which the intervals of input
variables [l  , l þ ] and [m  , m þ ] are divided into the small intervals to obtain Nl and Nm sample points. In the first stage
(Algorithm 1), starting from the lower bound of interval [l  , l þ ], the sample points are incremented from s¼1 until s¼Nl.
Each sample point ls is assumed as the centre point ln  1 of a new membership function (2). Assuming the desired poles
intervals (9), the controller parameters are calculated for the operating points (16) according to the equations system (12)
for an eliminant matrix Sk consisting of the crane dynamic model parameters Θ1 and Θ2 approximated through fuzzy
interpolation performed according to the formula (15).
ðln  1 ; m1 Þ; ðln  1 ; mr Þ; ð16Þ

It leads to complete the rules base by adding the two new rules:

RN þ 1 : If l is Ln  1 and m is M1 Then yN þ 1 is ½KTN þ 1 ; ΘT1;N þ 1 ; ΘT2;N þ 1 T

RN þ 2 : If l is Ln  1 and m is Mr Then yN þ 2 is ½KTN þ 2 ; ΘT1;N þ 2 ; ΘT2;N þ 2 T ð17Þ

where N is the number of fuzzy rules in the previous iteration.

 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81 75

In the next step, the condition (13) is tested for operating points corresponding to the crossover point of membership
functions μLn  1 ðlÞ and μLn  2 ðlÞ, and the centre points of functions μM1 ðmÞ and μMr ðmÞ:
ððln  1 þ ln  2 Þ=2; m1 Þ; ððln  1 þ ln  2 Þ=2; mr Þ: ð18Þ
If the condition (13) is satisfied, the temporary fuzzy set An  1 and rules (17) are removed from the fuzzy scheduler. If it is
violated, the new fuzzy set is created with the centre point of membership function at ls  1, and the conclusions of rules (17)
are determined for the operating point which has been tested successfully in the previous iteration.
Algorithm 1. Selecting the minimum number of fuzzy sets for l
Input: s: ¼1; n:¼ 2; N ¼ 4;
1: while s o Nl do
2: n:¼ nþ 1;
3: ln  1: ¼ls;
   þ 
4: define the fuzzy setLn  1 ¼ l; μLn  1 ðlÞ; l A l ; l , where μLn  1 ðlÞ is defined according to (2);
5: determineΘN þ 1 , ΘN þ 2 , [PN þ 1], [PN þ 2], KN þ 1, KN þ 2, for operating points (16) according to (12);
6: add the rules (17) to the RB;
7: N:¼ N þ2;
8: test the condition (13) for operating points (18);
9: if the condition (13) is satisfied
10: remove the fuzzy set Ln  1;
11: remove the rules (17) from the RB;
12: s: ¼s þ 1; n: ¼n-1; N:¼ N-2;
13: else
14: ln  1:¼ ls  1;
15: repeat step 5;
16: end if
17: end while

Algorithm 2. Selecting the minimum number of fuzzy sets for m

Input: s: ¼1; r:¼ 2; N¼ 2n;
1: while s o Nm do
2: r:¼ r þ1;
3: mr  1:¼ ms;
4: define the fuzzy setM r  1 ¼ fm; μMr  1 ðmÞ; m A ½m  ; m þ g, where μMr  1 ðmÞ is defined according to (3);
5: determine ΘN þ 1 , ΘN þ 2 ,…, ΘN þ n , [PN þ 1], [PN þ 2],…, [PN þ n], and KN þ 1, KN þ 2,…, KN þ n for operating points (20) according to (12);
6: add the rules (19) to the RB;
7: N:¼ N þn;
8: test the condition (13) for operating points (21);
9: if the condition (13) is satisfied
10: remove the fuzzy set Mr  1;
11: remove the rules (19) from the RB;
12: s: ¼s þ 1; r:¼ r-1; N: ¼N-n;
13: else
14: mr  1:¼ ms  1;
15: repeat step 5;
16: end if
17: end while

In the second stage, the Algorithm 2 leads to obtain the minimum number of fuzzy sets on the [m  , m þ ] universe of
discourse. At each iteration, the temporary fuzzy set Mr  1 is determined, and fuzzy rules are added to the current RB:
RN þ 1 : If l is L1 and m is Mr  1 Then yN þ 1 is ½KTN þ 1 ; ΘT1;N þ 1 ; ΘT2;N þ 1 T
RN þ 2 : If l is L2 and m is Mr  1 Then yN þ 2 is ½KTN þ 2 ; ΘT1;N þ 2 ; ΘT2;N þ 2 T
⋮
RN þ n : If l is Ln and m is Mr  1 Then yN þ n is ½KTN þ n ; ΘT1;N þ n ; ΘT2;N þ n T ð19Þ

where the conclusions parameters are determined for the operating points:
ðli ; mr  1 Þ f or i ¼ 1; 2; …; n: ð20Þ
The condition (13) is tested for most hazardous operating points associating with the crossover points of membership
functions:
ðli ; ðmr  1 þmr  2 Þ=2Þ f or i ¼ 1; 2; …; n;
ððli þ li þ 1 Þ=2; ðmr  1 þmr  2 Þ=2Þ f or i ¼ 1; 2; …; n  1; ð21Þ
ððli þ li þ 1 Þ=2; mr  1 Þ f or i ¼ 1; 2; …; n  1:

Owing to the assumed conditions, the two-stage procedure results in selecting the minimum number of membership
functions, their parameters distribution within the expected ranges of scheduling variables, as well as complete and
76 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81

coherent RB. Nevertheless, the parameters of the crane model should be identified at the new operating points determined
through applying Algorithm 1 and 2, and the iterative procedure should be repeated to validate the robustness of closed-
loop control system. After validation, the vectors Θ1k ¼ ½c0 ; d0 Tk (for k¼1, 2,…, N) are removed from rules conclusions to
obtain the fuzzy interpolation control scheme employed in the sensorless anti-sway control system presented in Fig. 1.
The proposed method of designing a fuzzy logic-based anti-sway crane control system involves to use the sway angle of a
payload sensor in open-loop experiments of crane dynamic model parameters identification at the operating points
corresponding to the bounds of scheduling variables ranges and selected through applying the iterative procedure.

4. Results of experiments conducted on the laboratory scaled overhead crane

The proposed algorithm was employed to design the fuzzy logic-based control scheme of a DC motors driven motion
mechanism of the laboratory scaled overhead crane. The hardware-software equipment utilized during identification and
control process was composed of PC with I/O board and Matlab software used in the open-loop identification experiments
and control system design, and PAC system with RX3i controller on which the control algorithm was implemented in the
form of structured text and tested during experiments conducted on the lab-scaled material handling device (Fig. 4). The
interval arithmetic-based procedure of the feedback control scheme synthesis was implemented in Matlab according to
Algorithms 1 and 2, thus the control system prototyping is not the time-consuming process which is carried out
automatically based on the data obtained from the open-loop identification experiments and resulting in the new operating
points determination, robustness of closed-loop system performances validation, and finally the parameters of a fuzzy
control scheme uploading to the data registers of RX3i controller.
The position and speed of crane, as well as the sway angle of payload sensing for identification purposes, were measured
using A/B phase incremental encoders. The sway angle of a payload can be measured using three methods elaborated on the
laboratory stand. The first method, which has been used in experiments presented in the paper, is based on the incremental
encoder used to measure the rotary angle of the fork-bottomed arm embracing a rope. In next technique, the Hall-effect
magnetic field sensors UV-3HF series are mounted on the hook assembly to measure its deflection from the equilibrium. The
last, non-contact method is based on the intelligent camera type of Sony XCI V3 installed under the trolley and directed to
the rope. In this technique, the vision software (Vision Builder for Automated Inspection) integrated with the embedded
system, allows to extract the specified rectangular region of snapshot with the isolated rope straight line, and determine the
angle between the detected rope edge and parallel line of image height. The vision-based measurement techniques
elaborated on the laboratory stand and results of experiments are detailed described in [30–32].
The objective of closed-loop system synthesis was to design a fuzzy scheduling control scheme taking into consideration the
parameters variation within the intervals [0.8, 2.2]m and [10,90] kg. The models of crane dynamic were identified at operating
points corresponding to the bounds of those intervals using output error (OE) method with sample time Ts ¼0.1 s. At each
operating point considering as the centre point of membership function, the intervals of desired stable poles were assumed as:
½zf k ¼ ½expðð ωn 8 0:1ωn ÞT s Þk ; ð22Þ

where the natural not dumped pulsation of the considered system varies in the interval ωn A ½2:06; 2:95rad/s.

Fig. 4. The laboratory stand: control system of the physical model of an overhead crane.
 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81 77

Fig. 5. The natural pulsation for operating points at which the parameters of a pendulum model were interpolated between points at which the OE
identification was conducted.

Through dividing the expected ranges of rope length and mass of a payload into Nl ¼14 and Nm ¼8 intervals, the iterative
method, which has been described in section 4, results in designing the fuzzy scheduler with 6 rules, and 3 and 2 fuzzy sets
determined for l and m, respectively, with the centre points fixed at l1 ¼0.8 m, l2 ¼ 1.5 m, l3 ¼ 2.2 m, and m1 ¼10 kg,
m2 ¼90 kg (Fig. 2). The open-loop identification experiments were conducted for operating points (1.5 m, 10 kg) and (1.5 m,
90 kg) determined using iterative method, and Algorithms 1 and 2 were used again to confirm the performances robustness
to fuzzy interpolation errors of closed-loop system parameters. Fig. 5 depicts the natural pulsation of a dynamic model at
the identified operating points and for operating points at which the model parameters were interpolated between bounds
of the interval [0.8, 2.2]m (during designing the fuzzy scheduler), and between bounds of the intervals [0.8, 1.5]m and [1.5,
2.2]m (during closed-loop control system validation). The maximum interpolation error was 7.26% in designing process, and
0.9% during validation.
Figures from 6 to 10 present the examples of unit step responses (crane position and the payload deviation from the
equilibrium position) obtained during experiments conducted for the operating points associating with the midpoints
(Figs. 6 and 7) and crossover points (Figs. 8–10) of triangular membership functions determined for the scheduling variables.
The response of the object, measured using the incremental encoder, was compared with the payload deviation estimated
based on the pendulum discrete-time model output applied in the sensorless feedback closed-loop anti-sway control
system (Fig. 1). The similarity of the pendulum model (7) and object responses in the transient states confirm assumptions
which were applied for object modeling.
The experiments conducted for the operating points varying within the intervals [0.8, 2.2] m and [10,90] kg confirmed
the robust performances: settle time, which depends on the natural pulsation of a crane system, was in the interval [5.5, 7] s,
while crane positioning and reduction of payload deviation were obtained with the expected tolerance 70.02 m. The same
results were obtained in the experiments realized with the feedback from sway angle of a payload sensor. Thus, the
robustness to fuzzy interpolation errors of closed-loop system parameters verified through repeating successfully the
iterative procedure (Algorithms 1 and 2) after identification of crane dynamic model parameters at operating points (1.5 m,
10 kg) and (1.5 m, 90 kg) was also confirmed by results of experiments, similar for the both approaches tested on the
laboratory stand: sensorless and sensor-based feedback control schemes. However, the sensorless approach is not
robustness in case of external disturbances affecting a crane-cable-payload system.
Fig. 11 presents the comparison of two experiments conducted for l ¼0.8 m and m ¼10 kg with using the fuzzy
scheduling control scheme and linear closed-loop control system (Fig. 3) designed for l ¼2.2 m and m¼ 90 kg and tested on
the laboratory stand with payload deflection feedback from the sensor of sway angle of a payload. In the first experiment,
the fuzzy interpolation-based control scheme places the closed-loop system characteristic polynomial coefficients at the
midpoint of desired interval determined based on the assumed interval of desired closed-loop system poles (22), that results
in payload positioning in the settle time about 5.5 s. In the second experiment, conducted for closed-loop system which
does not satisfy the condition (13) for the desired interval of poles (22) assumed for the operating point {0.8 m, 10 kg}, the
control system performances significantly deteriorate: although the maximum payload deflection is less, that is obviously
caused by the long rise time, it simultaneously results in worse settle time (about 9 s).

5. Conclusions

The fast, precise and safe transfer of goods in crane operations require a control application to solve the problems of
payload oscillations limitation. The open-loop control approaches are sensitive to the disturbances owing to the lack of sway
angle of a payload feedback. On the other hand, the closed-loop control methods require a reliable measurement system. In
78 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81

Fig. 6. Crane position and payload deviation – experiments for l ¼{0.8, 1.5, 2.2}m and m¼ 10 kg.

Fig. 7. Crane position and payload deviation - experiments for l ¼ {0.8, 1.5, 2.2}m and m¼90 kg.

spite of the fact, that the proposed method of designing an anti-sway crane control system involves sensing the payload
deflection in the open-loop identification experiments to estimate discrete-time model parameters, the anti-sway closed-
loop control scheme can be implemented in the applications which involve sensing the payload deflection (e.g. when the
 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81 79

Fig. 8. Crane position and payload deviation – experiments for l ¼{0.8, 1.5, 2.2}m and m¼ 50 kg.

Fig. 9. Crane position and payload deviation – experiment for l ¼ 1.15 m and m¼50 kg.

Fig. 10. Crane position and payload deviation - experiment for l ¼ 1.85 m and m ¼50 kg.

disturbances affect the closed-loop system performances) or in the sensorless approaches, when the disturbances can be
neglected and/or the reliable techniques of payload deflection sensing are costly to build and maintain. The sensorless anti-
sway discrete-time control approach is based on the feedback signal of payload deflection estimated by a pendulum model
formulated as the discrete-time relation between the sway angle of a payload and crane speed. This application was
successfully tested on the laboratory scaled overhead crane. The control system was built using PAC system with RX3i
80 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81

Fig. 11. Crane position and payload deviation – comparison of two experiments conducted for l¼ 0.8 m and m¼ 10 kg using the fuzzy scheduling control
scheme (payload deviation feedback from discrete-time model) and linear closed-loop control system designed for l ¼ 2.2 m and m¼ 90 kg (feedback from
the sensor of payload deflection).

controller. The results of experiments proved that the proposed fuzzy logic-based approach ensures robustness of closed-
loop system performances within the expected range of system parameters variation.
The papers describes the method of designing a fuzzy scheduler-based control scheme for a flexible dynamic system.
The most of the works reported in the literature relating to a fuzzy control of flexible dynamic system is based on the
linguistic rule-based strategies or a fuzzy controller designing through only tuning the membership functions and/or rules
conclusions parameters. The problem of fuzzy interpolation control scheme design is solved in this paper by interval
analysis of closed-loop control system characteristic polynomial coefficients. The proposed method, which has been based
on the interval Diophantine equation, is applied to find minimum number of fuzzy rules and fuzzy sets on the scheduling
variables universe of discourse and to determine the distribution of triangular-shaped membership functions parameters,
which satisfy the acceptable range of performances deterioration in the presence of system0 s parameters variation. The
future challenge is the implementation of this method for design the control system of a large scale material handling
device.

Acknowledgements

I thank the anonymous referees for their thorough review, comments and suggestions, that significantly contributed to
improve the quality of the work.

References

[1] Z.C. Qiu, Adaptive nonlinear vibration control of a Cartesian flexible manipulator driven by a ballscrew mechanism, Mech. Syst. Signal Process. 30
(2012) 248–266.
[2] Z.C. Qiu, B. Wang, X.M. Zhang, J.D. Han, Direct adaptive fuzzy control of a translating piezoelectric flexible manipulator driven by a pneumatic rodless
cylinder, Mech. Syst. Signal Process. 36 (2013) 290–316.
[3] Z. Smalko, J. Szpytko, Safety in engineering practice, in: Proceedings of 17th European Safety and Reliability Conference ESREL, Valencia, Spain, 2009,
pp. 1231–1237.
[4] J. Szpytko, D.A. Wozniak, To keep operational potential of transport device e-based on reliability indicators, in: Proceedings of European Safety and
Reliability Conference ESREL, Stavanger, Norway, 2007, pp. 2377–2384.
[5] M. Kenison, W. Singhose, Input shaper design for double pendulum planar gantry cranes, in: Proceedings of the IEEE Conference on Control
Applications, Hawaii, USA, 1999, pp. 539–544.
[6] J. Vaughan, A. Yano, W. Singhose, Comparison of robust input shapers, J. Sound Vib. 315 (2008) 797–815.
[7] X. Xie, J. Huang, Z. Liang, Vibration reduction for flexible systems by command smoothing, Mech. Syst. Signal Process. 39 (2013) 461–470.
[8] J.W. Auernig, H. Troger, Time optimal control of overhead cranes with hoisting of the load, Automatica 23 (4) (1987) 437–447.
[9] M. Volckaert, M. Diehl, J. Swevers, Generalization of norm optimal ILC for nonlinear systems with constraints, Mech. Syst. Signal Process. 39 (2013)
280–296.
[10] J. Huang, E. Maleki, W. Singhose, Dynamics and swing control of mobile boom cranes subject to wind disturbances, IET Control Theory Appl. 7 (9)
(2013) 1187–1195.
[11] A. Benhidjeb, G.L. Gissinger, Fuzzy control of an overhead crane performance comparison with classic control, Control Eng. Pract. 3 (12) (1995)
1687–1696.
[12] M. Mahfouf, C.H. Kee, M.F. Abbod, D.A. Linkens, Fuzzy logic-based anti-sway control design for overhead cranes, Neural Comput. Appl. 9 (2000) 38–43.
[13] J. Yi, N. Yubazaki, K. Hirota, Anti-swing and positioning control of overhead traveling crane, Inf. Sci. 155 (1–2) (2003) 19–42.
[14] X. Li, W. Yu, Anti-swing control for an overhead crane with fuzzy compensation, Intell. Autom. Soft Comput. 17 (X) (2004) 1–11.
[15] M.I. Solihin, A. Legowo Wahyudi, Fuzzy-tuned antiswing control of automatic gantry crane, J. Vib. Control 16 (1) (2010) 127–145.
[16] D. Liu, J. Yi, D. Zhao, W. Wang, Adaptive sliding mode fuzzy control for a two-dimensional overhead crane, Mechatronics 15 (5) (2005) 505–522.
[17] J. Smoczek, J. Szpytko, A mechatronics approach in intelligent control systems of the overhead traveling cranes prototyping, Inf. Technol. Control 37 (2)
(2008) 154–158.
[18] S.-K. Oh, W. Pedrycz, S.-B. Rho, T.-C. Ahn, Parameter estimation of fuzzy controller and its application to inverted pendulum, Eng. Appl. Artif. Intell. 17
(1) (2004) 37–60.
[19] N. Sadati, A. Hooshmand, Design of a gain-scheduling anti-sway controller for tower cranes using fuzzy clustering techniques, in: Proceedings of
International Conference on Computational Intelligence for Modeling, Control and Automation, Sydney, Australia, 2006.
 J. Smoczek / Mechanical Systems and Signal Processing 46 (2014) 70–81 81

[20] M.B. Trabia, J.M. Renno, K.A.F. Moustafa, Generalized design of an anti-swing fuzzy logic controller for an overhead crane with hoist, J. Vib. Control 14
(3) (2008) 319–346.
[21] C.-Y. Chang, Adaptive fuzzy controller of the overhead crane with nonlinear disturbances, IEEE Trans. Ind. Inf. 3 (2) (2007) 164–172.
[22] D. Liu, J. Yi, M. Tan, Proposal of GA-based two-stage fuzzy control of overhead crane, in: Proceedings of IEEE Conference on Computers,
Communications, Control Power Engineering 3, 2002, pp. 1721–1724.
[23] J. Smoczek, J. Szpytko, Design of gain scheduling anti-sway controller using genetic fuzzy system, in: Proceedings of 17th IFAC International
Conference on Methods and Models in Automation and Robotics MMAR, Miedzyzdroje, Poland, 2012, pp. 573–578.
[24] W. Yu, A. Moreno-Wrmendariz, F.O. Rodriguez, Stable adaptive compensation with fuzzy CMAC for an overhead crane, Inf. Sci. 181 (21) (2011)
4895–4907.
[25] A.Z. Al-Garni, K.A.F. Moustafa, S.S.A.K. Javeed Nizami, Optimal control of overhead cranes, Control Eng. Pract. 3 (1995) 1277–1284.
[26] R. Manning, J. Clement, D. Kim, W. Singhose, Dynamics and control of bridge cranes transporting distributed-mass payloads, J. Dyn. Syst. Meas. Control
132 (1) ;132http://dx.doi.org/10.1115/1.4000657.
[27] W. O0 Connor, H. Habibi, Gantry crane control of a double-pendulum distributed-mass load, using mechanical wave concepts, Mech. Sci. 4 (2013)
251–261.
[28] E.M. Abdel-Rahman, A.H. Nayfeh, Z.N. Masoud, Dynamics and control of cranes: A review, J Vib. Control 9 (2003) 863–908.
[29] R. Moore, Interval analysis, Prentice-Hall, Englewood Cliffs, NJ, 1966.
[30] P. Hyla, The crane control systems: a survey, 17th IFAC International Conference on Methods and Models in Automation and Robotics MMAR,
Miedzyzdroje, Poland, 2012 pp. 505–509.
[31] J. Smoczek, J. Szpytko, P. Hyla, Non-collision path planning of a payload in crane operating space, Solid State Phenom. 198 (2013) 559–564.
[32] J. Smoczek, J. Szpytko, P. Hyla, The anti-sway crane control system with using dynamic vision system, Solid State Phenom. 198 (2013) 589–593.