See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/41058005

Input shaping for vibration-free positioning of flexible systems

Article · January 2001

Source: OAI

CITATIONS READS
7 1,524

1 author:

Azdiana Md. Yusop

Technical University of Malaysia Malacca
28 PUBLICATIONS 151 CITATIONS

SEE PROFILE

All content following this page was uploaded by Azdiana Md. Yusop on 18 November 2015.

The user has requested enhancement of the downloaded file.

INPUT SHAPING FOR VIBRATION-FREE
POSITIONING OF FLEXIBLE SYSTEMS

AZDIANA BT. MD. YUSOP

UNIVERSITI TEKNOLOGI MALAYSIA

 “I hereby declare that I have read this thesis and
in my opinion this thesis is sufficient in terms of scope
and quality for the award of the degree of
Master of Electrical Engineering (Mechatronics and Automatic Control)”

Signature : _____________________
Supervisor : Dr. Zaharuddin bin Mohamed
Date : 13 November 2006
INPUT SHAPING FOR VIBRATION-FREE POSITIONING OF FLEXIBLE
SYSTEMS

AZDIANA BINTI MD. YUSOP

A thesis submitted in fulfillment of the

requirements for the award of the degree of
Master of Electrical (Mechatronics and Automatic Control) Engineering

Faculty of Electrical Engineering

Universiti Teknologi Malaysia

NOVEMBER 2006
 ii

I declare that this thesis entitled “Input Shaping for Vibration-Free Positioning of
Flexible Systems “ is the result of my own research except as cited in the references. The
thesis has not been accepted for any degree and is not concurrently submitted in
candidature of any other degree.

Signature: …………………………

Name: AZDIANA BT. MD. YUSOP

Date: 13 November 2006

 iii

To my beloved mom and dad

 iv

ACKNOWLEDGEMENT

Firstly, I would like to thank my supervisor, Dr Zaharuddin bin Mohamed, for all
his teachings and guidance, his criticism on me and my work, and his experience help on
this work. It has been a hardship for you, sorry and thank you so much. I would also
like to thank all other colleagues and friends for their help, discussions and information
sharing. Finally thank you my father and mother for all your love, sacrifice,
understanding and support, without which, I could never ever walk the first step.
 v

ABSTRACT

Input shaping is a simple method for reducing the residual vibration in

positioning lightly damped systems. For controlling part, a continuous and
differentiable function is introduced to define the desired motion and the input is shaped
by inverse dynamic analysis. The shaped input function is derived from the specified
output function so that the designer can choose the speed and shape of the motion within
the limitations of the drive system. Third order exponential function will be used as the
desired output due to its asymptotic behavior. The simulation has been done to the
spring-mass-damper system which is a second order system to study the application of
the technique to the system. The effects of errors in damping ratio and natural frequency
are also discussed. Next, the same technique is applied to a gantry crane system which
is fourth order system. In the proposed method the parameters that need to be defined is
the position of the trolley and sway angle of the mass. Simulated responses of the
position of the trolley and sway angle of the mass are presented using MATLAB. The
performance of the Bang-bang input technique and the inverse dynamic analysis are
compared. From the simulation results, satisfactory vibration reduction of a crane
system has been achieved using the proposed method.
 vi

ABSTRAK

‘Input shaping’ merupakan kaedah mudah untuk mengurangkan getaran semasa

menggerakkan sesuatu sistem. Pada bahagian pengawal, fungsi persamaan yang
berterusan dan boleh beza diperkenalkan untuk mendapatkan respons yang dikehendaki
dan persamaan input diterbitkan menggunakan teknik ‘inverse dynamic’. Persamaan
input diperolehi daripada respons output yang dikehendaki supaya pengkaji dapat
memilih kelajuan dan bentuk respons yang diperlukan supaya berada dalam had
maksima sesuatu sistem. Fungsi eksponen kuasa tiga akan digunakan sebagai output
disebabkan oleh sifat kestabilan asimptotnya. Simulasi dijalankan ke atas sistem spring-
beban teredam iaitu sistem order kedua untuk mengkaji kesan teknik ini kepada sistem
tersebut. Kesan ralat pada ‘damping ratio’ dan ‘natural frequency’ juga dibincangkan.
Seterusnya, teknik yang sama diaplikasikan kepada sistem kren ‘gantry’ yang
merupakan sistem order keempat. Dengan menggunakan teknik ini, parameter yang
akan dikaji adalah kedudukan troli dan sudut ayunan beban. Respons bagi kedudukan
troli dan sudut ayunan beban akan ditunjukkan menggunakan perisian MATLAB.
Prestasi output menggunakan input Bang-bang dan ‘inverse dynamic’ dibandingkan.
Dari keputusan simulasi didapati pengurangan kadar getaran yang memuaskan telah
diperolehi menggunakan teknik yang diperkenalkan.
 vii

TABLE OF CONTENTS

CHAPTER TITLE PAGE

DECLARATION ii
DEDICATION iii
ACKNOWLEDGEMENT iv
ABSTRACT v
ABSTRAK vi
TABLE OF CONTENTS vii
LIST OF TABLES x
LIST OF FIGURES xi
LIST OF SYMBOLS xiii

I INTRODUCTION 1
1.1 Project Introduction 1
1.2 Objective 3
1.3 Methodology 3
1.4 Project Overview 4
1.5 Thesis Outline 4
 viii

II LITERATURE REVIEW 5
2.1 Introduction 5
2.2 Posicast Control 5
2.3 Command Shaping 6
2.4 Convolution (Impulse Shaping) 7
2.5 Bang-bang Control 9
2.6 Zero Vibration 10
2.7 Zero Vibration and Derivative 10
2.8 Extra Insensitive 11
2.9 Time-Optimal Rigid-Body (TORB) Command 12
and Time-Optimal Flexible-Body (TOFB)
2.10 Time-Delayed Control 13
2.11 System Inversion Based Method 14

III SYSTEM INVERSION BASED METHOD 15

3.1 Introduction 15
3.2 Inverse Dynamic 15
3.3 Desired Motion 18
3.4 Required Shaped Input 20

IV MODELLING OF THE GANTRY CRANE SYSTEM 23

4.1 Introduction 23
4.2 Model Description 25
4.3 Derivation of the Equations of Motion 26
4.4 Linearization 34
4.5 Inverse Dynamic Analysis 36
4.6 Representation in Time Domain 38
4.7 Summary 41
 ix

V SIMULATION RESULTS AND ANALYSIS 43

5.1 Introduction 43
5.2 MATLAB and SIMULINK 44
5.3 SIMULINK Model of the Spring-Mass- 45
Damped System (second order system)
5.4 SIMULINK Model of the Gantry Crane
System (4th order system) 53
5.4.1 Generation of Bang-bang Input Force 54

VI CONCLUSION 63
6.1 Conclusion 63
6.2 Future Works 65
6.2.1 Experimental Setup 65
6.2.2. Design of the Closed-loop Control 65
System

REFERENCES 66
 x

LIST OF TABLES

TABLE NO. TITLE PAGE

3.1 Characteristics of the proposed output function 19

5.1 Required shaped input and output response for 48

various β values

5.2 Effect of errors in ξ on the system response 52

5.3 Effect of errors in wn on the system response 53

5.4 Time response of trolley position 62

 xi

LIST OF FIGURES

FIGURE NO. TITLE PAGE

2.1 Posicast control 6

2.2 Multi pulse shaped input 7
2.3 An example of impulse shaping technique 8
2.4 Impulse shaper versus step shaper 9
2.5 Input shaper 10
2.6 Sensitivity curve 11
3.1 System model used in the examples 16
3.2 Design process 22
4.1 Model of a Gantry crane 25
5.1 Simulation model using MATLAB with step input function 45
5.2 Output response of the system with step input function 45
5.3 Simulation model with input shaping 46
5.4 System response to simplified input frequencies for various β 50
values
5.5 Effect of errors in ξ ±20% (β=2) 51
5.6 Effect of errors in ξ ±40% (β=2) 51
5.7 Effect of errors in wn ±20% (α=2.15) 52
5.8 Effect of errors in wn ±40% (α=2.15) 53
5.9 Generation of Bang-bang input force 54
5.10 Parameters of the Bang-bang input 54
5.11 General model 55
5.12 Nonlinear model 56
 xii

5.13 Sway motion when input force is positive 57

5.14 Sway motion when input force is negative 57
5.15 Input plot 59
5.16 Sway angle 59
5.17 Arc length and chord length 60
5.18 Trolley position 61
 xiii

LIST OF SYMBOLS

ωn - Natural frequency
ξ - Damping ratio
K - Stiffness
u - Normalized time
α - Speed motion
β - Relationship between α and ωn
M - Trolley mass
m - Payload mass
l - Length of the hoisting rope
Fx - Input force
g - Gravitational acceleration = 9.81ms-2
G - Centre point
S - Point of suspension
x - Trolley position
x& - Velocity
&x& - Acceleration
θ - Sway angle
θ& - Angular velocity
&θ& - Angular acceleration
 CHAPTER I

INTRODUCTION

1.1 Project Introduction

In many machines, load positioning is achieved by simple open-loop control. In

the case where structural flexibility is significant, and the load is lightly damped, the
vibration may be unacceptable. Many solutions have been proposed to reduce vibration
using input shaping technique.

Vibration is a serious problem in mechanical systems that are required to

perform precise motion in the presence of structural flexibility. Examples of such
systems range from the positioning of disk drives head to large space structures, flexible
manipulators and container cranes. In most cases, the residual vibration at the end of a
move is the most detrimental and the extent of the residual vibration limits the
performance of the system. The effective use of such systems can only be achieved
when such vibration can be properly handled. As a result, there is active research
interest in finding methods that will eliminate vibration for a variety of mechanical and
structural systems.
 2

Traditional closed-loop feedback can be used to reduce end-point vibration. The

closed-loop system will then benefit from the inherent advantages of feedback, such as
insensitivity to parameter variations, noise attenuation and disturbance rejection.
However, such a feedback system can be difficult to implement in practice, as it requires
reliable sensor information for feedback. Such sensor information may not be so easily
available. For example, in the container crane system problem, it is not a trivial task
(nor practical due to reliability of sensors and its environment) to devise a sensor to
measure the position at the end-point. Another approach is input shaping technique, in
which the input is preshaped such that the resulting residual vibration is reduced or
eliminated. These methods are popular in industry because they are relatively simple to
implement the preshaped input together with closed-loop feedback strategies to enjoy
the benefits of both systems.

Input shaping is a feedforward control technique for improving the settling time
and the positioning accuracy, while minimizing residual vibrations, of computer-
controlled machines. Input shaping is a strategy for the generation of time-optimal
shaped commands using only a simple model, which consists of the estimates of natural
frequencies and damping ratios. Hence, input shaping is a simple method for reducing
the residual vibration when positioning lightly damped systems. It offers several clear
advantages over conventional approaches for trajectory generation:

i) Designing an input shaping does not require an analytical model of the system; it
can be generated from simple, empirical measurements of the actual physical
system.

ii) Input shaping does not affect the stability of the closed loop system in any way.
It simply modifies the command signal to the system so that all moves,
regardless of length, are vibration free.
 3

1.2 Objective

This project attempts to specify an input function that will drive the system from
an initial position into a target position as fast as possible without vibration at the target
position and within the physical constraints of the drive system.

1.3 Methodology

i) Study on the inverse dynamic analysis to shape the input function of the system.

ii) Derive the shaped input function from the specified output function, in this case
is a third order exponential function.

iii) Implement the input function into the open loop system.

iv) Develop the dynamic model using MATLAB and SIMULINK.

v) Investigating of the technique to a gantry crane system.

 4

1.4 Project Overview

As a whole, this project considers only one parameter that needs to be defined
that is output speed, which is limited only by the physical constraints of the drive
system. A continuous and differentiable function is introduced to define the desired
motion and the input is shaped by inverse dynamic analysis. The system output function
is specified and the shaped input function will be derived. Third order exponential
function is used as the desired output due to its asymptotic behavior. Simulation is done
using MATLAB to obtain the output response. From the simulation results, under
certain circumstances, the design process can be simplified and the need for inverse
dynamics is eliminated. In addition, robustness is evaluated by a sensitivity analysis on
the simulated examples.

1.5 Thesis Outline

This thesis consists of six chapters. Chapter I provides some background of the
project, the objective and the scope of studies. Chapter II contains the literature review
on several important concepts of input shaping, technology and tools used in the study.
Chapter III entails the principle of system inversion based method including the
behaviour of the method on second order system. Chapter IV follows with the design
and modelling of the gantry crane system. Simulation results, analysis and discussion of
the performance of the technique are presented in Chapter V. The work is then
concluded in Chapter VI with some suggestions and future works.
 CHAPTER II

LITERATURE REVIEW

2.1 Introduction

Precise position control and rapid rest-to-rest motion is the desired objective in a
variety of applications. The requirement of precise position control implies that the
residual vibration of the structure should be zero or near zero. There are several
techniques of input shaping that will reduce vibration and is discussed below.

2.2 Posicast Control

An early form of input shaping was the use of Posicast control by O. J. M. Smith
(1958). The control motion is like casting a fly in Figure 2.1, hence the name positive-
cast, or Posicast, control. This technique breaks a step input into two smaller steps, one
of which is delayed in time [1]. The result is a reduced settling time for the system.
 6

Figure 2.1: Posicast Control

2.3 Command Shaping

Between the late 50’s and the publication of the Input Shaping paper by Singer
and Seering [2], there was some work on the shaping of input profiles for control of
residual vibration [3], [4]. Swigert [4] proposed techniques for the determination of
torque profiles which considered the sensitivity of the terminal states to variations in the
model parameters. Publication of the Input Shaping paper renewed interest in
prefiltering reference inputs for robust vibration control, which has resulted in dozens of
papers with application to spacecrafts, robots, cranes, chemical processes, etc.

The design will consider two classes of problems. The first, involves real-time
shaping or time-delay filtering of the reference command to stable systems with the
objective of minimizing the residual vibration [5]. The second class of problems
considered is the design of controllers for systems with rigid body modes with
constraints on the control input [6], [7]. The example of this technique can be seen in
Figure 2.2.
 7

Figure 2.2: Multi Pulse Shaped Input

2.4 Convolution (Impulse Shaping)

The most popular technique for input shaping is to convolve a sequence of

impulses and various methods for shaping impulse function and examining their
robustness has been reported and applied to flexible spacecraft, robots and to the control
of swing of suspended objects transported by cranes [8]. Impulse shaping is a
feedforward control technique for reducing vibrations in computer controlled machines.
The method works by creating a command signal that cancels its own vibration. That is,
vibration caused by the first part of the command signal is canceled by vibration caused
by the second part of the command.

Input shaping is implemented by convolving a sequence of impulses, an input

shaper, with any desired command. The shaped command that results from the
convolution is then used to drive the system. See Figure 2.3. If the impulses in the
 8

shaper are chosen correctly, then the system will respond without vibration to any
unshaped command.

Figure 2.3: An example of impulse shaping technique

The amplitudes and time locations of the impulses are obtained from the system's
natural frequencies and damping ratios. Shaping can be made very robust to errors in
the system parameters. The impulse sequence is chosen such that in the absence of
control input, it itself would not cause residual vibration. Recall that convolution in the
time domain is equivalent to multiplication in the Laplace domain. In order to increase
the rise time, when using impulse shaping, the impulses are allowed to take negative
value [9] and multihump shaping of the impulses can be used to increase the system
robustness [10].

A step shaper is designed by generating acceleration/deceleration pulse profile

using step inputs with alternative sign. The main difference between impulse shaper and
step shaper is whether it uses the convolution concept or not. Figure 2.4 shows the
impulse shaper versus the step shaper.
 9

Figure 2.4: Impulse shaper versus step shaper

2.5 Bang-bang Control

Other methods proposed by Meckl and Seering consist of using a multiswitch

bang–bang forcing function, which gives time-optimal performance [11], or adding up
harmonics of ramped sinusoid functions in order to approximate as close as possible a
bang–bang function, but minimizing the energy introduced at system resonance
frequencies.
 10

2.6 Zero Vibration

A zero vibration (ZV) shaper is designed by requiring that it produce commands

that cause zero residual vibration when the model is perfect. For a single-mode system,
this constraint leads to a shaper containing two impulses as shown in Figure 2.5.
However, the ZV shaper is very sensitive to modeling errors.

Figure 2.5: Input shaper

2.7 Zero Vibration and Derivative

A zero vibration and derivative (ZVD) shaper is obtained by requiring the

derivative with respect to the frequency of the residual vibration be equal to zero. That
is, the sensitivity curve must have zero slopes at the modeling frequency. It is shown
that ZVD shaper provides higher robustness than the ZV shaper. Moreover, higher
robustness can further be achieved by higher derivatives.
 11

2.8 Extra Insensitive

Unlike the ZV and ZVD shapers, the extra insensitive (EI) shaper does not
attempt to force the vibration to zero at the modeling frequency. Rather, the vibration is
limited to some low, but acceptable level of residual vibration. The sensitivity curve for
an EI shaper designed to limit vibration below 5% is shown in Figure 2.6.

Subsequent developments of this approach are the two-humped and three-

humped EI [12]. In general, the increasing of robustness of the system (obtained with a
larger number of impulses adopted in the shaper) is paid with an increasing of system
delays, due to the convolution process.

Figure2.6: Sensitivity Curve

 12

2.9 Time-Optimal Rigid-Body (TORB) Command and Time-Optimal Flexible-

Body (TOFB)

The efficiency of cargo handling work at a port depends largely on the operation
of container cranes. When a ship is unloaded, containers are first transferred from the
ship to a waiting truck by a container crane. The truck then carries the container to an
open storage area, where another crane stacks the container to a pre-assigned place. The
bottleneck of this cycle lies in the transfer of the containers from the ship to the truck.

Therefore, minimizing this transfer time will bring about a large cost saving.
When a ship is loaded, the same problem is encountered. Since a large swing of the
container load during the transfer is dangerous, the problem is to transfer a container to
the desired place as quickly as possible while minimizing the swing of the container
during transfer as well as the swing at the end of transfer.

If the oscillation of the container load is ignored, TORB commands can be easily
calculated. Unfortunately, TORB commands will usually result in large amplitude
oscillations. Experienced crane operators attempt to eliminate vibration by causing a
deceleration oscillation which cancels the oscillation induced during acceleration, or
they may brush the payload against obstacles to damp out the vibration.

When the swing is considered, TOFB commands that result in zero residual
vibration can be generated. Hoisting of the load during the motion increases the
difficulty of generating the control because the system is nonlinear. If the system model
is linearized, then the associated frequency is time-varying. Optimal controls based on a
nonlinear model can be difficult to generate. One method for developing optimal
controls divides the motion into fundamental parts. The control for each part is then
 13

derived and pieced together. Even when optimal commands can be generated,
implementation is usually impractical because the boundary conditions at the end of the
maneuver (move distance) must be known at the start of the move. When feedback is
available, both robust controllers and combination of open- and closed-loop controls are
possible.

2.10 Time-Delayed Control

This method uses the direct and time-delayed signal to cancel the poles of a
system with the intention of attenuating the residual vibration. Robust Time-Delayed
Control is the method referred to as the proportional plus multiple delay control,
involves the use of multiple time delays in conjunction with a proportional part to cancel
the dynamics of the system in a robust fashion.

All the above methods start with a parametric input function, which usually
involves magnitude and time delay. The parameter values are calculated in order to
reduce the residual vibrations at the final position. The speed of the system is
determined mainly by the system dynamics and little control can be exercised on the
speed of the response. In all cases, the achievement of robustness or the control of
vibration leads to an increase in system delays.
 14

2.11 System Inversion Based Method

A system inversion based method has been suggested recently by Piazzi and
Visioli [13] and compared with bang-bang and other impulse shaping input methods.
This proposed method are discussed and illustrated by simulations and experiments on
the next chapter.
 CHAPTER III

SYSTEM INVERSION BASED METHOD

3.1 Introduction

System inversion based method reverse the process by specifying the system
output function and deriving the input. In this way, the designer can choose the speed
and shape of the motion within the limitations of the drive system. In the present study,
in order to control the motion after the end-point, it is necessary to use output function
with only one parameter and this leads to a much simpler input function. The single
parameter of this output function determines the motion speed and is limited only by the
drive system constraints.

3.2 Inverse Dynamic

Consider a lightly damped second-order system with damping ratio ξ, natural

frequency ωn and output displacement x(t). The objective is to specify an input function
that will drive the system from an initial position (0) into a target position, XE as fast as
 16

possible without vibration at the target position and within the physical constraints of the
drive system. In this case, the value of XE is assumed to be 1.

M
f(t)

x(t)

Figure 3.1: System model used in the examples

If the input is force, f(t) as shown in Figure 3.1 and K is the stiffness, then the equation
of motion can be written as

dx(t ) d 2 x(t )
f (t ) = Kx (t ) + C +M (3.1)
dt dt 2
f (t ) = Kx (t ) + Cx& (t ) + M&x&(t ) (3.2)

Using the Laplace transform, Equation (3.2) can be obtained as:

F ( s ) = KX ( s ) + CsX ( s ) + Ms 2 X ( s ) (3.3)

F ( s ) = X ( s )[ K + Cs + Ms 2 ]
X (s) 1
G ( s) = = (3.4)
F ( s ) K + Cs + Ms 2
 17

Equation (3.4) is comparing the transfer function of the system in Figure 3.1 with the
second order G(s) transfer function:

X ( s) ωn 2

G ( s) = = 2
F ( s) s + 2ζω n s + ω n 2

yields;

K C
ωn 2 = , 2ζω n =
M M

From Equation(3.2);

C K f (t )
&x&(t ) + x& (t ) + x(t ) = (3.5)
M M M

From the comparison that have been made:

&x&(t ) + 2ζω n x& (t ) + ω n 2 x(t ) = ω n f (t ) / k (3.6)

For the example shown in Figure 3.1, if the desired output of the system together with
the first and second order derivatives as a function of time can be specified, then these
expressions can be inserted into the left hand side of Equation (3.6) to obtain the
necessary input function to produce the desired output.
 18

3.3 Desired Motion

The function representing the point-to point motion must be differentiable and
continuous at least up to second order derivatives where the velocity and acceleration at
the start and at the arrival of the destination are zero and remain zero. Asymptotic
behavior is desired when approaching the final position in order to stay there and
exponential functions is the best solution. The third-order exponential function for the
output motion is examined to move the system from zero to XE;

3
x(t ) = X E (1 − e − (αt ) ) (3.7)

In order to generalize the analysis, normalized time u is defined as:

u = αt (3.8)

This gives the following normalized equations of the desired motion:

x(t ) 3
X (u ) = = 1 − e −u (3.9)
XE

The derivatives are:

1 3 d 3
X& (u ) = x& (t ) = −e −u ( −u 3 ) = 3u 2 e −u (3.10)
αX E du

1 d −u 3 3 d 2 3
X&& (u ) = &x&(t ) = 3[u 2 (e ) + e −u (u )] = (6u − 9u 4 )e −u (3.11)
α XE
2
du du

Table 3.1 shows the characteristics of the proposed output function defining by
displacement, velocity and acceleration.
 19

Displacement

x(t ) 3
X (u ) = = 1 − e −u
XE

Displacement (m)
Normalised time, u=αt

Velocity

1
X& (u ) = x& (t )
αX E
3 d
= −e −u (−u 3 )
Velocity (m/s)
du
3
= 3u 2 e −u

Normalised time, u=αt

Acceleration

1
X&& (u ) = &x&(t )
α XE 2
)

d −u 3 3 d
2

= 3[u 2 (e ) + e − u
Acceleration (m/s

(u 2 )]
du du
3
= (6u − 9u 4 )e −u

Normalised time, u=αt

Table 3.1: Characteristics of the proposed output function

 20

From Table 3.1, the maximum of normalized output velocity is 1.18 and the
maximum value of the normalized output acceleration is 2.15. Therefore, if there are
restrictions on the output velocity or acceleration, then from Equation (3.10) the
maximum value of α can be obtained as follows:

max X&
α max = (3.12)
1.18 X E

max X&&
α max = (3.13)
2.15 X E

This value gives the minimum settling time achievable in terms of the limits of the
physical values of velocity or acceleration. Equation (3.12) is used if there is any
limitation in velocity whereas Equation (3.13) will be taken into account when
acceleration is limited. Obviously, if both velocity and acceleration are limited, then the
smaller of the two α max values calculated in both equations should be used.

3.4 Required Shaped input

In order to obtain the required input to achieve the desired motion, the
expressions for the output displacement, velocity and acceleration in Equations (3.9),
(3.10) and (3.11) are substituted in Equation (3.6):

α 2 X&& (u ) X E + 2ζω nαX& (u ) X E + ω n 2 X (u ) X E = ω n 2 f (t ) / k

2ζω n & ω ω f (t )
2 2

X&& (u ) + X (u ) + n2 X (u ) = n
α α KX E α 2

X&& (u ) + 2ζβ X& (u ) + β 2 F (u ) = β 2 F (u ) (3.14)

 21

Where:
ωn 1
β= and F (u ) = f (t )
α KX E

β determines the relationship between the system speed (ωn) and the speed motion (α).
In physical terms, large values of β define a motion that is slow relative to the system
natural frequency.

By inserting Equations (3.9), (3.10) and (3.11) into Equation (3.6), the
normalized input function to produce the desired motion can be solved:

3 3 3
(6u − 9u 4 )e − u + 2ζβ (3u 2 e − u ) + β 2 (1 − e − u ) = β 2 F (u )
3
(6u − 9u 4 + 6ζβ u 2 − β 2 )e − u + β 2
F (u ) =
β2

3 ⎛ 6u − 9u 4 6ζu 2 ⎞ −u 3
F (u ) = (1 − e −u ) + ⎜⎜ + ⎟⎟e
⎝ β
2
β ⎠

⎛ 6u − 9u 4 6ζu 2 ⎞ −u 3
F (u ) = X (u ) + ⎜⎜ + ⎟⎟e (3.15)
⎝ β 2
β ⎠

First Second
term term

The input force expression consists of two terms, first term is the required output
and the second term will approaches zero in the steady state, but it compensates some of
the system dynamics in order to achieve the required motion. The second term is
examined more clearly in the simulation results.
 22

Figure 3.2 shows the design process of the system inversion based method
technique.

Start

Generate
desired motion

Calculate α, β

NO Use simplified
β<8 Input function

YES
Calculate input by
inverse dynamic

Implement the input

function into the
open loop system

Obtain the output

response

Stop

Figure 3.2: Design process

 CHAPTER IV

MODELLING OF THE GANTRY CRANE SYSTEM

4.1 Introduction

A gantry crane system is a crane carrying the trolley or trolley with a movable or
fixed hoisting mechanism, that the bridge is rigidly supported on two or more legs
running on fixed rails or other runway. The fundamental motions of a gantry crane
consist of traversing, load hosting and load lowering. Gantry cranes are widely used as
an efficient means of traversing heavy object in factories, warehouse and shipping yards.
Like other crane types, gantry cranes met with some dissatisfactory due to its natural
characteristics.

As mentioned, the fundamental motions of a gantry crane consist of traversing,

load hosting and load lowering. These significant characteristic is that all motions are
performed simultaneous at relatively high speed. Crane traversing motions, particularly
when starting or stopping; induce undesirable swinging of the suspended load. This
creates another problem that the swing could cause the hosting rope to leave its groove
which could lead to over wrapping and damage.
 24

One of the characteristics of these cranes is the flexible hoisting ropes used as a
part of the structure for the reduction of system mass, which result in favorable features
of high payload ratio, high motion speed and low power consumption. However, the
flexible hoisting create serious problems, that is the crane acceleration which required
for motion will generate undesirable load swing, which is frequently aggravated by load
hoisting. Therefore, such load swing should be suppressed as rapidly as possible to
maximize the operations

The operation of the gantry crane systems in many industrial settings is achieved
by relying on the skill of experienced crane operators. Unfortunately, precise payload
positioning, meaning that the operator using only visual feedback to position the
payload, is difficult due to the reality that the payload is free to swing in a pendulum like
motion. In addition, the payload swing can result in several performances and safety
concerns that include damage to the payload like spillage or breakage, damage to the
surrounding environment of personnel, and large internal forces that can result in
reduced payload carrying capacity or premature failure of stressed part.

This chapter will emphasize on the modeling of a gantry crane and the
application of the inverse dynamic analysis into the system. The system model will
then be represented in time domain and in state-space form.
 25

4.2 Model Description

The model of a gantry crane is shown in Figure 4.1. Generally, the

configuration of the gantry crane model is specified by the horizontal position of the
trolley, x, the length of the hosting rope, l, and the swing angle of the rope, θ. The
payload, which is suspended from the point of suspension, S, is assumed to be a rigid
body symmetric about its axis with mass m and centre point , G of mass m.

y axis

Trolley
x axis ro M
Point of suspension, S

rm ym
θ l

xm m
Centre point, G
x Payload

mg

Figure 4.1: Model of a Gantry Crane

Before the derivation of the equations of motion, some assumptions are made
for simplicity. Firstly, friction force that may exist in the trolley is to be ignored. The
trolley and the payload can be considered as point masses. Besides, the tension force
 26

that may cause the hoisting rope elongate is also neglected. The trolley and the payload
are assumed to move in x-y plane, which means a study of two dimensional.

4.3 Derivation of the Equations of Motion

To study the dynamic behaviors of the gantry crane model, the position vector
of point of suspension, S and the centre point G with respected to the fixed axes
coordinates have to be determined. As shown in Figure 4.1, the position vector of S
with respected to the fixed axes coordinate system is

~ ~
ro = xi + yj (4.1)
~
= xi

~ ~
where i and j are x and y directional unit vectors, respectively.

The position vector of G with respected to the fixed axes coordinate system is
related by the moving circular coordinate, θ, which can be written as

~ ~
rm = xmi + ymj (4.2)

where xm = x + l sin θ and ym = −l cos θ

Therefore, the kinetic energy of the system T can be obtained as:

 27

T = Ttrolley + Tpayload
1 1
= Mr&o 2 + mr&m 2
2 2
1 1
(
= Mx& 2 + m x& m 2 + y& m 2
2 2
) (4.3)

where M and m are the mass of the trolley and the payload, respectively.

The potential energy of the system, P can be represented as:

P = mgym
= − mgl cos θ (4.4)

where g is the gravitational acceleration or g = 9.81ms −2 .

Using the Lagrangian function, the following equation is derived as:

L =T−P
1
2
1
( )
= Mx& 2 + m x& m 2 + y& m 2 + mgl cos θ
2
(4.5)

From above, it is noticed that xm = x + l cos θ and ym = −l cos θ . Hence,

the first derivative of xm and ym can be obtained as:

dxm
x& m =
dt
d(x + l sin θ)
=
dt
= x& + l&sin θ + θ& (l cos θ)
= x& + sin θl& + l cos θθ&
 28

And

dym
y& m =
dt
d(− l cos θ )
=
dt
[
= − l& cos θ + θ& l(− sin θ)]
= l sin θθ& − l& cos θ

Therefore,

( 2
) (
xm 2 + ym 2 = x& + sin θl& + l cos θθ& + l sin θθ& − l&cos θ
2
)
+ (sin θl& + l cos θθ& ) + 2x& (sin θl& + l cos θθ& ) + l sin
2
= x& 2 2 2
θθ& 2
+ l&2 cos 2 θ − 2ll&sin θ cos θθ&
(
= x& 2 + sin 2 θl&2 + l2 cos 2 θθ& 2 + 2ll&sin θ cos θθ& + 2x& sin θl& + l cos θθ& )
+ l2 sin 2 θθ& 2 + l&2 cos 2 θ − 2ll&sin θ cos θθ&
( ) ( )
= x& 2 + l&2 sin 2 θ + cos 2 θ + θ& 2l2 sin 2 θ + cos 2 θ + 2x& sin θl& + 2x& l cos θθ&
= x& 2 + l&2 + l2θ& 2 + 2x& l&sin θ + 2x& lθ& cos θ

By substituting xm 2 + ym 2 into equation (4.5), the Lagrangian function is

obtained as

L=
1
2
1
(
Mx& 2 + m x& 2 + l&2 + l 2 θ& 2 + 2 x& l&sin θ + 2x& lθ& cos θ
2
) (4.6)
+ mgl cos θ

After obtaining the Lagrangian function, L, the equations of motion can be

derived. The Lagrange’s equation is defined as:

d⎛ ∂ ⎞ ∂
⎜ ⎟− = Qi
dt ⎜⎝ ∂q& ⎟⎠ ∂q i = 1, 2 …… r (4.7)
 29

Thus, the equations of motion associated with the generalized coordinates

q = [x, θ, l] are represented.
T

First, the relation of the.2 dimensional forces is formulated with q = x, which is:

d ⎛ ∂L ⎞ ∂L
⎜ ⎟− = Fx
dt ⎝ ∂x& ⎠ ∂x

∂L
=0
∂x

∂L
∂x&
1
(
= Mx& + m 2 x& + 2l&sin θ + 2lθ& cos θ
2
)
= Mx& + mx& + ml&sin θ + mlθ& cos θ

d ⎛ ∂L ⎞
(
⎜ ⎟ = M&x& + m&x& + m l&&sin θ + l&θ& cos θ +
dt ⎝ ∂x& ⎠
)
[
m l&θ& cos θ + l&θ& cos θ + lθ& (− sin θ)θ& ]
(
= (M + m )&x& + m l&&sin θ + l&θ& cos θ + )
(
m l&θ& cos θ + l&θ& cos θ − lθ& 2 sin θ)
Thus, (
= (M + m )&x& + ml &θ& cos θ − θ& 2 sin θ + )
2ml&θ& cos θ + ml&&sin θ

d ⎛ ∂L ⎞
⎜ ⎟ − 0 = Fx
dt ⎝ ∂x& ⎠
 30

And

(
Fx = (M + m )&x& + ml &θ& cos θ − θ& 2 sin θ +)
2ml&θ& cos θ + ml&&sin θ (4.8)

Next, let the generalized coordinate q = θ and the following equation can be
derived:

d ⎛ ∂L ⎞ ∂L
⎜ ⎟− =0
dt ⎝ ∂θ& ⎠ ∂θ

∂L 1
[ ]
= m 2x& l& cos θ + 2x& lθ& (− sin θ) + mgl(− sin θ )
∂θ 2
1
( )
= m 2x& l& cos θ − 2x& lθ& sin θ − mgl sin θ
2

∂L 1
(
= m 2l 2 θ& + 2 x& l cos θ )
∂θ& 2

d ⎛ ∂L ⎞ 1
[ ]
⎜ ⎟ = m 2l &θ& + 2l&θ& (2l) + 2&x&l cos θ + 2 x& l& cos θ + 2x& lθ& (− sin θ)
dt ⎝ ∂θ& ⎠ 2
2

1
( )
= m 2l 2 &θ& + 4ll&θ& + 2&x&l cos θ + 2x& l& cos θ − 2 x& lθ& sin θ
2
Since

d ⎛ ∂L ⎞ ∂L
⎜ ⎟− =0
dt ⎝ ∂θ& ⎠ ∂θ
 31

Then

d ⎛ ∂L ⎞ ∂L
⎜ ⎟=
dt ⎝ ∂θ& ⎠ ∂θ
2l2&θ& + 4ll&θ& + 2&x&l cos θ + 2x& l&cos θ − 2 x& lθ& sin θ = 2x& l&cos θ − 2x& lθ& sin θ − gl sin θ
2l2&θ& + 4ll&θ& + 2&x&l cos θ = −2gl sin θ
2l&θ& + 4l&θ& + 2&x& cos θ + 2g sin θ = 0

And

(4.9)
l&θ& + 2l&θ& + &x& cos θ + g sin θ = 0

Now, let the q = l, the relation of the 2 dimensional forces is formulated as the
following:

d ⎛ ∂L ⎞ ∂L
⎜ ⎟− = Fl
dt ⎝ ∂l& ⎠ ∂l

∂L 1
( )
= m 2lθ& 2 + 2 x& θ& cos θ + mg cos θ
∂l 2

∂L 1
&
(
= m 2l& + 2 x& sin θ
∂l 2
)

d ⎛ ∂L ⎞ 1
(
⎜ ⎟ = m 2l&& + 2&x& sin θ + 2x& θ& cos θ
dt ⎝ ∂l& ⎠ 2
)
 32

Since,

d ⎛ ∂L ⎞ ∂L
⎜ ⎟− = Fl
dt ⎝ ∂l& ⎠ ∂l

Then,

Fl =
1
2
( )
m 2l&& + 2&x& sin θ + 2 x& θ& cos θ

⎡1
( )
− ⎢ m 2lθ& 2 + 2 x& θ& cos θ + mg cos θ⎥
⎤
⎣2 ⎦
= ml&& + m&x& sin θ + mx& θ& cos θ − mlθ& 2 − mx& θ& cos θ − mg cos θ

And

(4.10)
Fl = ml&& + m&x& sin θ − mlθ& 2 − mg cos θ

Thus, the equations of motion of the gantry crane model associated with the
generalized coordinates q = [x, θ, l] can be summarized, respectively, as:
T

x:
( )
Fx = (M + m )&x& + ml &θ& cos θ − θ& 2 sin θ +
2ml&θ& cos θ + ml&&sin θ

θ: l&θ& + 2l&θ& + &x& cos θ + g sin θ = 0

l: Fl = ml&& + m&x& sin θ − mlθ& 2 − mg cos θ

 33

The above nonlinear dynamic model equations (4.8) – (4.10) can be represented
in the following matrix form:

M (q)q
&& + C(q& , q)q& + g (q) = F (4.11)

where state vector q and driving force vector F are defined as q = [x, θ, l]
T
and

F = [Fx ,0, Fl] , respectively. M(q) is the 3 × 3 symmetric mass matrix and C(q& ,q) is the
T

3× 3 Coriolis and centrifugal forces matrix and g(q) is the 3 × 1 gravitational loading
vector.

Therefore, the M(q) can be written as

⎡ M + m ml cos θ m sin θ⎤
M (q) = ⎢⎢ cos θ l 0 ⎥⎥ (4.12)
⎢⎣m sin θ 0 m ⎥⎦

and C(q,
& q) can be written as:

⎡0 ml& cos θ − θ& sin θ mθ& cos θ⎤

⎢ ⎥
C(q& ,q) = ⎢0 l& θ& ⎥
(4.13)
⎢0 − m l &
θ 0 ⎥
⎣ ⎦

and the g(q) is:

⎡ 0 ⎤
⎢
g (q) = ⎢ g sin θ ⎥⎥ (4.14)
⎢⎣− mg cos θ⎥⎦
 34

Hence, the representation of the dynamic behaviors of the model in matrix form
is:

⎡Fx ⎤ ⎡ M + m ml cos θ m sin θ⎤ ⎡&x& ⎤

⎢ 0 ⎥ = ⎢ cos θ l 0 ⎥⎥ ⎢⎢&θ& ⎥⎥ +
⎢ ⎥ ⎢
⎢⎣ Fl ⎥⎦ ⎢⎣m sin θ 0 m ⎥⎦ ⎢⎣l&&⎥⎦
(4.15)
⎡0 ml&cos θ − θ& sin θ mθ& cos θ⎤ ⎡ x& ⎤ ⎡ 0 ⎤
⎢ ⎥⎢ & ⎥ ⎢
⎢0 l& θ ⎥ ⎢ θ ⎥ + ⎢ g sin θ ⎥⎥
&
⎢0 − mlθ& 0 ⎥⎦ ⎢⎣ l& ⎥⎦ ⎢⎣− mg cos θ⎥⎦
⎣

4.4 Linearization

The derived model which is represented in matrix form is a nonlinear model.

Therefore, for a better progress of modeling, the nonlinear dynamic model has to be
linearized.

The equations of motions are nonlinear is due to the presence of the

trigonometric terms such as sin θ and cos θ as well as the quadratic term θ& 2 . Hence,
to do the linearization, the Taylor series expansion is used:

f (x) = f (x0) +
df (x − x0) + d 2 f (x − x0)2 +K
dx 1! dx 2 2! (4.16)
x =x0 x=x0

or

df
f (x) ≈ f (x0) + (x − x0)
dx x =x0
(4.17)
 35

therefore,

sin θ ≈ sin θ0 + cos θ0 (θ − θ0 )

Assume the swing angle, θ is sufficiently small, hence

θ0 = 0

and

sin θ ≈ θ

Similar to the trigonometric term cos θ , therefore

cos θ ≈ cos θ0 + sin θ0 (θ − θ0 )

and

cos θ ≈ 1

Apart from that, the θ& is assumed sufficiently small. So the quadratic term θ& 2 is
negligible, which is:

θ& 2 ≈ 0

Since the tension force that may cause the hoisting rope elongate is neglected,
therefore the length of the hoisting rope is assumed to be constant, which is:

l& = l&& = 0
 36

Thus, for the generalized coordinates q = [x, θ, l] , the linearized model of a

T

gantry crane with the equations motion is obtained as:

x: (4.18)
Fx = (M + m )&x& + ml&θ&

θ: (4.19)
l&θ& + &x& + gθ = 0

l: (4.20)
Fl = m&xθ
& − mg

Generally, θ is very small of its value. This makes Fl = m&x&θ − mg ≈ − mg . This

suggests that the hoisting motion of the gantry crane is decoupled from the trolley if
such approximation is made. It makes that only equations (4.18) and (4.19) valid for
deriving input in inversion method technique.

4.5 Inverse Dynamic Analysis

The linearized equation will be used to derive the input function using inverse
dynamic analysis. The same third-order exponential function for the output motion is
examined to move the system from zero to XE;

3
x(t ) = X E (1 − e − (αt ) ) (4.21)

Equation (4.18) can be divided into two terms as:

 37

f (t ) = ( M + m) &x&(t ) + mlθ&&(t ) (4.22)

First Second
term term

In order to generalize the analysis, normalized time u is defined as:

u = αt (4.23)

This gives the following normalized equations of the desired motion:

x(t ) 3
X (u ) = = 1 − e −u (4.24)
XE

The derivatives are:

1 3 d 3
X& (u ) = x& (t ) = −e −u (−u 3 ) = 3u 2 e −u (4.25)
αX E du

1 d −u 3 3 d 2 3
X&& (u ) = &x&(t ) = 3[u 2 (e ) + e − u (u )] = (6u − 9u 4 )e −u (4.26)
α XE
2
du du

By inserting the normalized equation and all the derivates, equation (4.22) can be
obtained as;

F (u ) = ( M + m)α 1 X&& (u ) + mlα 2 θ&&(u )

2 2
(4.27)

F1 (u) F2 (u)

3
F1 (u ) = ( M + m)α 1 (6u − 9u 4 )e − u
2
(4.28)
 38

F2 (u ) = 0 (4.29)

As the sway angle ( θ ) is required to be zero or no vibration, equation F2(u) is assumed

to be zero because the second term of Equation (4.22) consists of term θ&&(t ) .

Therefore Equation (4.22) can be written as:

3
F (u ) = ( M + m)α 1 (6u − 9u 4 )e − u
2
(4.30)

Equation (4.30) is known as the input function and will be applied into the nonlinear
gantry crane system and the results will be discussed in the next chapter.

4.6 Representation in Time Domain

After getting the linearized equation, these equations can be represented in time
domain or in state-space representation.

A state-variable representation corresponding to equations (4.18) and (4.19) is

obtained by defining the state vector as:

[x, θ, x& , θ& ]T

Then,
 39

dx dθ
x& = θ& =
dt dt

dx& &
&x& = &θ& = dθ
dt dt

From equation (4.19),

&x& = −l&θ& − gθ

Substituting &x& to equation (4.18) yields,

(M + m )(− l&θ& − gθ ) + ml&θ&

Fx =

Fx = −(M + m )l&θ& − (M + m )gθ + ml&θ&

Fx = &θ&[ml - (M + m )l] − (M + m )gθ
Fx = &θ&(− Ml) − (M + m )gθ

Thus,

&θ&(Ml) = −(M + m )gθ − Fx

&θ& = −⎛⎜ M + m ⎞⎟gθ − Fx

⎝ Ml ⎠ Ml
&θ& = − ⎡⎛⎜ M + m ⎞⎟gθ + Fx ⎤ (4.31)
⎢ Ml Ml ⎥⎦
⎣⎝ ⎠

Substituting &θ& to equation (4.18) yields,

Fx (M + m )&x& − ml⎡⎢⎛⎜ M + m ⎞⎟gθ + Fx ⎤

Ml ⎥⎦
=
⎣⎝ Ml ⎠
⎛M+m⎞ ⎛ Fx ⎞
Fx = (M + m )&x& − ml⎜ ⎟gθ − ml⎜ ⎟
⎝ Ml ⎠ ⎝ Ml ⎠
 40

Thus,

(M + m )&x& = Fx + m⎛⎜ Fx ⎞⎟ + m⎛⎜ M + m ⎞⎟gθ

⎝M⎠ ⎝ M ⎠

(M + m )&x& = Fx ⎜⎜⎜ M + m ⎟⎟⎟ + m⎛⎜ M + m ⎞⎟gθ

⎛ ⎞
(4.32)
⎝ M ⎠ ⎝ M ⎠
⎛m⎞
&x& = Fx + ⎜ ⎟gθ
M ⎝M⎠

For the representation in state-space, the four equations can be put into a
standard matrix form as:

x& = Ax + Bu

with

⎡0 0 1 0⎤
⎢0 0 0 1⎥⎥
⎢ mg
A = ⎢0 0 0⎥
⎢ M ⎥
⎢ (M + m )g ⎥
⎢⎣0 − Ml
0 0⎥
⎦

and

⎡ 0 ⎤
⎢ 0 ⎥
⎢ 1 ⎥
B=⎢ ⎥
⎢ M ⎥
⎢ 1 ⎥
⎢⎣− Ml ⎥⎦
 41

Thus,

⎡0 0 1 0⎤ ⎡ 0 ⎤
⎡ x& ⎤ ⎢ ⎡x ⎤ ⎢
⎢ θ& ⎥ ⎢0 0 0 1⎥ ⎢ ⎥ ⎢ 0 ⎥⎥
⎥
⎢ ⎥ = ⎢0 mg θ 1
0 0⎥ ⎢ ⎥ + ⎢ ⎥ Fx (4.33)
⎢&x& ⎥ ⎢ M ⎥ ⎢ x& ⎥ ⎢ M ⎥
⎢&& ⎥ ⎢ (M + m )g ⎥⎢ ⎥ ⎢ 1 ⎥
0 0⎥ ⎣ θ& ⎦ ⎢−
⎣ θ ⎦ ⎢0 −
⎣ Ml ⎦ ⎣ Ml ⎥⎦

The outputs of interest are:

y = Cx + Du

Therefore,

⎡x ⎤
⎢ ⎥
⎡ θ ⎤ ⎡0 1 0 0 ⎤ ⎢ θ ⎥
=
⎢ x ⎥ ⎢1 0 0 0⎥ ⎢ x& ⎥
⎣ ⎦ ⎣ ⎦ (4.34)
⎢&⎥
⎣θ⎦

By obtaining the dynamic model in time domain and frequency domain,

simulation and analysis of the gantry crane can be performed.

4.7 Summary

This chapter is mainly deal with the derivation of equations of motion of a gantry
crane. To achieve that, a model of a gantry crane is introduced with parameters such as
 42

trolley mass, payload mass, length of the hoisting rope as well as the force. The
concerned issue of the gantry system is mainly the sway angle and its trolley position.

The equations of motion are derived using Lagrangian approach. Since the
system is nonlinear, thus, it is being linearized by using some assumptions to eliminate
some nonlinearities of the system. Using the linearized equation of motion, the input
function is derived using inverse dynamic analysis.
 CHAPTER V

SIMULATION RESULTS AND ANALYSIS

5.1 Introduction

The contribution of the inverse dynamic analysis has been investigated by

implement the input function into a spring-mass-damper system and a gantry crane
system The dynamic of the systems has been discussed earlier in the previous chapter.

This chapter will describe the MATLAB and SIMULINK which is used to
simulate the dynamic behaviors, the simulation results and the system analysis on the
both system. The results presented in this thesis are based on the design procedure that
has been stated before. In the last section of this chapter, the discussion of the results is
also described.
 44

5.2 MATLAB and SIMULINK

MATLAB is a technical computing environment for high performance numeric

computation and visualization. MATLAB integrates numerical analysis, matrix
computation, signal processing and graphics in an easy-to-use environment where
problems and solutions are expressed just as they written mathematically without
traditional programming.

MATLAB can extend functionally and versatility with addition of optional

application -specific toolboxes. Toolboxes are comprised of suites of MATLAB
function (M-files) written by world-class authorities on each of the respective topic.
MATLAB is an open environment for which many specialized toolboxes have been
developed: Control System, Signal Processing, Optimization, Robust Control, Micro-
analysis and Synthesis (micro-tools), System Identification, MMLE State Space
Identification and Neural Network.

Simulink (formerly known as Simulab) is a graphical environment for the

modeling and simulating block diagram and general nonlinear systems. It is a
companion program to MATLAB and an interactive system for simulating nonlinear
dynamic system. It is a graphical mouse-driven program that allows to model a system
by drawing a block diagram on the screen and manipulating it dynamically. It can work
with linear, nonlinear, continuous, discrete-time, multivariable and multi rate systems.

Block sets are add-ins to Simulink that provide additional libraries of blocks for
specialized applications like communications, signal processing, and power systems.
Real-time workshop is a program that allows to generate C code from block diagrams
and to run it on a variety of real-time system.
 45

5.3 SIMULINK Model of the Spring-Mass-Damped System (second order

system)

The process is demonstrated by the following example applied to the single-

degree of freedom system as shown in Figure 5.1 with the parameters set to M=1 kg,
K=800 N/m, C=9 N s/m, ωn=28.28 rad/s, ξ=0.159.

Figure 5.1: Simulation model using MATLAB with step input function

Figure 5.1 shows the simulation model of second order system. With a step
input function, it will cause vibration about 60% overshoot as shown in Figure 5.2.
Output (m)

Time (sec)

Figure 5.2: Output response of the system with step input function
 46

The inverse dynamic input shaping technique has been applied to the system in
Figure 5.3 in order to reduce the vibration. Table 5.1 shows the responses of the input
function, F(u) and output response X(u) with various values of β. The second term is
also plotted (dotted lines) to illustrate the contribution of this term.

Figure 5.3: Simulation model with input shaping

 47

F(u) X(u)
β=1.0

Input (m)

Output (m)
Time (sec) Time (sec)

β=2.0

Output (m)
Input (m)

Time (sec) Time (sec)

β=5.0
Input (m)

Output (m)

Time (sec) Time (sec)

 48

F(u) X(u)
β=6.0

Input (m)

Output (m)
Time (sec) Time (sec)

β=8.0

Output (m)
Input (m)

Time (sec) Time (sec)

β=10.0
Input (m)

Output (m)

Time (sec)
Time (sec)

Table 5.1: Required shaped input and output response for various β values :
contribution of the second term in Equation (3.14)
 49

From the simulation results in Table 5.1, for low values of β (fast motion),
second term dominates the input and the system requires high input values (force or
torque) in order to achieve the desired motion. However, at a larger value of β, the
second term is not significant and oscillations disappear.

In applications where the system does not require to move as fast as β<8 owing
to limitations of drive system, system can be simplified because the input signal is the
same as the desired output displacement i.e.:

3
F (u ) = X (u ) = 1 − e − u (5.1)

Simulations of the normalized system equation were carried out with the
simplified input function and the results are shown in Figure 5.4 for different values of
β. The omission of the second term did not cause any problem for β>8. By using the
simplified input function, there may be slight changes in the maximum acceleration and
velocity of the output which have to be taken into account if the speed of motion is
maximized by using the constraints on the velocity and acceleration.
 50

Output (m)

Time (sec)

Figure 5.4: System response to simplified input frequencies for various β values.

5.3.1 Robustness

A sensitivity analysis was performed to see the effect of errors in the natural
frequency and the damping ratio on the system response. In this work, ±20% and ±40%
errors in natural frequency and damping ratio have been taken into consideration.

Figures 5.5 and 5.6 show that the effect of the error in natural frequency is
significant when compared with the desired output. From Table 5.2, overestimation and
underestimation of errors in damping ratio will results in increasing settling time, Ts.
 51

The vibration occurs and become larger when the percentage of error increases.
Therefore the comparison of the desired output and the system actual output should
reveal the direction of corrections needed in the parameter estimation.

Output (m)

Time (sec)

Figure 5.5: Effect of errors in ξ ±20% (β=2)

Output (m)

Time (sec)

Figure 5.6: Effect of errors in ξ ±40% (β=2)

 52

% of error 20% of error in ξ 40% of error in ξ

Damping ratio ξ=0.1272 ξ=0.159 ξ=0.1908 ξ=0.0954 ξ=0.159 ξ=0.2226
Settling time, Ts 0.14 0.12 0.24 0.36 0.12 0.46
Overshoot (%) 2.34 0 3.31 4.67 0 7
Table 5.2: Effect of errors in ξ on the system response.

The value of β plays an important role in the input shaping process and gives and
indication of the motion speed with reference to the system natural frequency. From
Figures 5.7 and 5.8, it can be seen that by underestimating wn means that the actual β
value is smaller than that suppose to use, and care should be taken to avoid it.
Obviously, overestimation of wn will result in slower motion of response than it can be
achieved. In addition, the settling time is decreased when errors occur in natural
frequency. as presented in Table 5.3.
Output (m)

Time (sec)

Figure 5.7: Effect of errors in wn ±20% (α=2.15)

 53

Output (m)

Time (sec)

Figure 5.8: Effect of errors in wn ±40% (α=2.15)

% of error 20% error in wn 40% error in wn

Natural frequency wn wn wn wn wn wn
=22.624 =28.28 =33.936 =16.968 =28.28 =39.592
Settling time, Ts 0.74 0.744 0.72 0.74 0.744 0.72
Table 5.3: Effect of errors in wn on the system response.

5.4 SIMULINK Model of the Gantry Crane System (4th order system)

The performance of the system using Bang-bang input and system inversion
method is being compared through the simulation results.
 54

5.4.1 Generation of Bang-bang Input Force

The bang-bang input force which is applied to the trolley is generated as the
model below and is shown in the scope input. Figure 5.10 shows the parameters of the
Bang-bang input force that must be taken into account to fulfill the same output
characteristic as using the inverse dynamic analysis. In this case, the trolley is required
to move from zero position to final position that is 1 meter.

Figure 5.9: Generation of Bang-bang Input Force

Figure 5.10: Parameters of the Bang-bang input

 55

5.4.2 General Model

The general model is an overall system that is build to show the simulation from
diversity perspective which comprising of nonlinear system, linear system, state-space
model and transfer function of the system (frequency domain). The simulation is mainly
on the trolley position and the sway angle. Figures 5.11 and 5.12 show the models of
the gantry crane developed within SIMULINK.

Figure 5.11: General Model

 56

FigureFigure 5.2: General

5.12: Nonlinear Model
Model
 57

5.4.3 Sway Motion of a Gantry Crane

Figures 5.13 and 5.14 show the sway motion of the payload when an input force
is applied to the trolley. The motion of the payload is a pendulum like motion where the
payload will swing from its initial position to a final position. By definition, the initial
position has an angle of zero. When the input force is positive, as shown in Figure 5.13,
the sway angle will be a negative value, which swings clockwise by definition. Vice
versa, when the applied force is negative, the sway angle, which swings anticlockwise,
will be positive in value, as shown in Figure 5.14. From the perspective of the
transformation of energy, the initial position point has a maximum value of kinetic
energy whereas the final position point has the maximum potential energy.

Fx Fx

-θ θ

Figure 5.13: Sway Motion when Figure 5.14: Sway Motion when
Input Force is Positive Input Force is Negative

From the following simulation results, which the bang-bang torque is applied to
the trolley, It can be seen that the sway angle in radian have a trend related to the applied
force directions. However, when the applied force is taken off (Fx = 0N), it is found that
the load still oscillate and is having a large oscillation. This result indicates that the
payload motion has a very significant response to acceleration or deceleration
commands without any feedback control.
 58

5.4.4 Sway Angle

Basically, small load swing implies small crane acceleration. Accordingly, the
constraints of small swing, slow hoisting speed, and small hoisting distance are the
major sources of reducing the productivity and efficiency of crane operation.

In this work, the following gantry crane parameters are used. The parameters
correspond to the experimental crane system presented by Yong-Seok Kim, Han-suk
Seo and Seung-Ki Sul in A New Anti-sway Control Scheme for Trolley Crane System
[15].

• Trolley mass, M = 1 kg
• Payload mass, m = 0.8 kg
• Length of the hoisting rope, l = 0.305m

and the bang-bang input torque is assumed to be 1.8Nm. Figure 5.15 shows simulated
bang-bang input force and input function of the inverse dynamic analysis, whereas
Figure 5.16 shows the sway angle of the gantry crane system using both techniques.
 59

BANG-BANG INPUT INVERSE DYNAMIC ANALYSIS

Figure 5.15: Input plot

BANG-BANG INPUT INVERSE DYNAMIC ANALYSIS

Peak-to-peak =0.038rad
Peak-to-peak =0.28 rad

Figure 5.16: Sway angle

 60

It is noted that the motion of the payload is affected by the applied force which it
will swing according to its path defined earlier. In bang-bang input, the payload
continues to oscillate although the force is taken off by the time 2s. The payload
oscillates with a maximum value 0.28 rad, which is approximately 16.04 degree from its
initial position. This value is become significant when the length of hoisting rope
become larger. This is due to the chord length, which is nearly equal to the arc length
that proportionally increases with the length of the hoisting rope. See Figure 5.17:

θ
Hoisting rope
length

Chord length
Arc length

Figure 5.17: Arc Length and Chord Length

Thus, from the simulation result, the 0.28 rad will create a 8.4 cm chord length,
which is considered as a large swing in this case. For industrial use gantry crane, this
will cause nearly 1m chord length which will create safety problems and damages to the
surrounding environment.

Hence, by using inverse dynamic technique, the payload oscillate with a

maximum value 0.038 rad, which is approximately 2.18 degree from its initial position
It will only create a 1.1 cm chord length, resulting the vibration reduction of sway angle
in this technique is 86.4% compared to bang-bang input.
 61

5.4.4 Position of the Trolley

The simulation results of the position of the trolley with the bang-bang input and
inverse dynamic are shown in Figure 5.18:

BANG-BANG INPUT INVERSE DYNAMIC ANALYSIS

Figure 5.18: Trolley position

From the above graph, it can be clearly seen that the vibration occur when using
Bang-bang input. It reaches about 1.22m at 2sec. The inverse dynamic analysis
provides no vibration during positioning the trolley along its path and reach 1.22m at
3sec where the force is taken off. This is the dissatisfactory of this technique where its
performance is slower than bang-bang input force. The time response of the trolley
position has been simplified in Table 5.4. Although the settling time and rise time of
Bang-bang input force is faster than inverse dynamic technique, in terms of vibrations,
the performance of inverse dynamic technique is better than bang-bang input force.
 62

Bang-bang input Inverse dynamic technique

Settling time (sec) 2 3
Rise time (sec) 1.274 1.52
Table 5.4: Time response of trolley position
 CHAPTER VI

CONCLUSION

In this chapter, a conclusion for the thesis and suggestion for future work
concerning the improvements and upgrading of the gantry crane dynamics behaviors is
presented.

6.1 Conclusion

Generally, this project is mainly to specify an input function that will drive the
system from an initial position into a target position as fast as possible without vibration
at the target position and within the physical constraints of the drive system. The
application of inverse dynamic method to reduce vibration on a spring-mass-damper
system and into a gantry crane system has been investigated.
 64

Instead of selecting a suitable form for the input function and shaping it to
eliminate system vibrations, a suitable form for the output motion is developed, and then
the input is shaped to produce the desired output by inverse dynamics. The desired
output function is based on a third order exponential function, which satisfies all of the
requirements for system positioning.

By using the developed model, the dynamics behaviors are being simulated using
MATLAB and SIMULINK. The dynamic model and the results are discussed.
Simulation results have shown that an inversion technique can be applied to control
vibration. Satisfactory vibration reduction of a crane system has been achieved using
the inverse dynamic technique.

Basically, in gantry crane system, the motion of the payload and the trolley is
unstable with the occurrence of oscillation. When the applied force is taken off, the
system is undamped and seems to be swinging on their varying frequencies. These
situations are affected by the parameters such as the weight of the trolley, weight of the
payload and the length of the hoisting rope which affect the swinging frequency of the
system. The applied force, as the control unit of the system affects the system especially
on the swinging magnitude and the traveling distance.

The developed model is generally unstable. This requires controlling part in

order to reduce the oscillation of the payload. It is not just for the safety issue, which the
great oscillation will cause damages to the environments and human around, but at the
same time to transfer the payload at fast speed with less residual oscillation. This may
bring the idea towards future work on designing the control system for the gantry crane.
 65

6.2 Future Works

6.2.1 Experimental Setup

One of the limitations of this project is lacking out of the experimental setup.
This laboratory is a pilot model which may be a scaled model of an actual gantry crane
system. It is used mainly to compare the actual sway angle with estimated sway angle.

6.2.2 Design of the Closed-loop Control System

The design of the crane control system is the future work that can be the next
target to reduce the sway angle of the payload. Several closed-loop techniques can be
investigated including linear control, adaptive control, fuzzy logic control and nonlinear
control.
 66

REFERENCES

1. Smith, O. J. M., “Posicast Control of Damped Oscillatory Systems”, Proc. of the

IRE, 1957, pp 1249-1255.

2. Singer, N. C., and Seering, W. P., “Preshaping Command Inputs to Reduce

System Vibrations”, ASME J. of Dynamic Systems, Measurement and Control,
Vol. 112, 1990, pp 76-82.

3. Farrenkopf, R. L., 1979, “Optimal Open-Loop Maneuver Profiles for Flexible

Spacecraft”, J. of Guidance, and Control, Vol. 2, No. 6, pp 491-498

4. Swigert, C. J., “Shaped Torques Techniques”, J. of Guidance and Control, Vol.

3, 1980, pp 460-467.

5. Singh, T. and Vadali, S. R., 1993, “Robust Time-Delay Control”, ASME J. of

Dynamic Systems, Measurement,and Control, Vol. 115, pp. 303-306.

6. Singh, T. and Vadali, S. R., “Robust Time-Optimal Control: Frequency Domain

Approach”, AIAA J. of Guidance,Control and Dynamics, Vol. 17, No. 2, 1994,
pp 346-353.

7. Pao, L. Y. and Singhose, W. E., “Robust Minimum Time Control of Flexible

Structures”, Automatica, 34(2): 229-236, Feb. 1998.
 67

8. Cho, J. K. and Park, Y. S., “Vibration Reduction in Flexible Systems Using a

Time-Varying Impulse Sequence”, Robotica, 1995, 13, 305-313.

9. Singhose, W. E., Seering, W. P. and Singer, N. C., “Time-Optimal Negative

Input Shapes”, Trans. ASME, J. Dynamic Syst., Measmt, and Control, 1997, 119,
198-205.

10. Singhose, W. E., Porter, L.J., Tuttle, T. D. and Singer, N. C., “Vibration
Reduction Using Multi-Hump Input Shapers”, Trans. ASME, J. Dynamic Syst.,
Measmt, and Control, 1997, 119, 321-326.

11. P. Meckl andW. Seering, “Active damping in a three-axis robotic manipulator,”

ASME J. Vibr., Acoust., Stress, Reliab. Des., vol. 107, pp. 38–46, Jan. 1985.

12. W. E. Singhose, L. J. Porter, T. D. Tuttle, and N. C. Singer, “Vibration reduction

using multi-hump input shapers,” ASME J. Dynam. Syst., Meas.,Contr., vol. 119,
pp. 320–326, June 1997.

13. Piazzi, A. and Visioli, A., “Minimum Time System Inversion-Based Motion
Planning for Residual Vibration Reduction”,IEEE/ASME Trans. Mechatronics,
2000, 5(1), 12-22.

14. E. M. Abdel-Rahman Et Al (2003), “Dynamics and Control of Cranes: A

Review”, Journal of Vibration and Control 9: 863-908, 2003.

View publication stats