MODERN AND INTELLIGENT CONTROLLER FOR A

MAGNETIC BEARING SYSTEM

SHARATUL IZAH BINTI SAMSUDIN

A thesis submitted in fulfilment of the

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

Faculty of Electrical Engineering

Universiti Teknologi Malaysia

NOVEMBER 2006
 ABSTRACT

A magnetic bearing system is a device that uses electromagnetic forces to

support a rotor without mechanical contact. The focus of this project will be on the
stability and control of the MBC 500 system test bed constructed by Magnetic
Moments Incorporated. The MBC 500 system contains a stainless steel shaft or
rotor, which can be levitated using eight horseshoe electromagnets, four at each end
of the rotor. A controller, which is able to stabilize the position of the rotor by
varying the electromagnet force, 4 produced by the electromagnets at the end of the
shaft, will be designed. For this purpose, the formulation of the mathematical
dynamic model of magnetic bearing system is derived initially and it was followed
by establishing the state space model of the system. Then, system model is
linearized at the equilibrium point using a Tdylor Series and the shaft is assumed as a
rigid body. In addition, a state feedback controller using a pole placement technique
and a fUzzy logic controller as an altetnative control strategy are designed. This
project will be implemented using M A ~ A B
6.5.
 Sistem magnetik bering adalah satu perkakasan yang menggunakan daya
elektromagnetik untuk menyokong rotor tanpa memerlukan aplikasi mekanikal.
Fokus utama projek ini adalah pada kestabilan dan pengawalan sistem MBC 500
yang dibina oleh Magntic Moments Incorporated. Sistem MBC 500 ini meliputi aci
tahan karat atau rotor, yang mana boleh diapungkan menggunakan lapan ladam
elektromagnet, di mana terdapat empat ladam elektromagnet pada setiap hujung
rotor. Satu pengawal direkabentuk untuk menstabilkan kedudukan rotor dengan
mengubah daya electromagnet, 4 yang dihasilkan pada hujung aci. Untuk tujuan ini,
model matematik dinarnik bagi sistem magnetik bering ini dimmuskan pada awalnya
dan kemudian disusuli dengan model keadaan-mang bagi sistem ini. Seterusnya,
model sistem ini dilinearkan pada titik keseimbangan dengan menggunakan Siri
Taylor sementata aci dianggap sebagai badan tegar. Selain daripada itu, satu
pengawal suapbklik keadaan yang menggunakan teknik "pole-placement" berserta
pengawal ''fbzzy logic" sebagai pengawal alternatif ditekabentuk. Projek ini
dijalankan dengan menggunakan perisian MATLAB 6.5.
 CHAPTER 1

INTRODUCTION

1.1 Project Ovemew

Magnetic bearing is a device that uses electromagnetic forces to support a

rotor without mechanical contact. Magnetic bearings can be divided into two
categories which are passive and active magnetic bearing. Passive magnetic bearings
typically use permanent magnets in conjunction with electromagnets. With
permanent magnets, the force exerted on the rotor can be either attractive or
repulsive. A repulsive force results in a system that is stable without a controller.
However, the force exerted by the permanent magnets cannot be controlled and it is
limited by the strength of the magnets. On the other hand, for active magnetic
bearing, the force on the rotor can be controlled by changing the current flow in the
magnet coils [9]. The problem of using an active magnetic bearing is that it can only
exert an attractive force and make the system inherently unstable [14] and requiring
the use of a controller.
13 Scopes of Project

This project presents a study of designing a controller for magnetic bearing

system based on the following:

1. Formulation and proving the mathematical dynamic model of magnetic

bearing system.
2. The design of modem controller which is able to stabilize the position of
the rotor during operation. For this task, the state feedback control using
pole placement technique is applied.
3. The design of intelligent controller as to maintain the stability of the rotor
of a magnetic bearing system.

This project will be focused on designing a controller for MBC 500 magnetic
bearing system.

1.4 Research Methodology

The research work is undertaken in the following eight developmental stages:

1. Prove the mathematical dynamic model of a magnetic bearing system.

2. Establish the state space model of a magnetic bearing system.
3. Linearization: Nonlinear equations of a magnetic bearing system are
linearized at the equilibrium point using a Taylor Series.
4. Check the controllability and observability of a magnetic bearing system.
5. Design a state feedback controller using the pole placement technique.
6. Design an intelligent controller as an alternative control strategy.
 7. Verify and analyse the controller design of a magnetic bearing system
simulated on MATLAB-SIMULINK.
8. Evaluate the results as to compare the performance of both controller
stated.

1.5 Literature Research

The research on stabilizing and controlling a rotor of a magnetic bearing

system has gained momentum over the last decade. This is due to the nonlinear and
inherently unstable dynamics of the system. As the applications for active magnetic
bearing can be found widely, the importance for designing the appropriate and
efficient controller to monitor the magnetic bearings becomes vital. The following
paragraphs briefly discuss on several researches that have been done by researchers.

MBC 500 magnetic bearing system has been identified in designing a

classical controller and this was done by J. Shi and J. Revell (2002). MATLAB p -
Analysis and Synthesis toolbox was applied in system identification. Hewlett
Packard 3562A Dynamic Signal Analyzer is used to collect the experimental data
fiom the MBC 500 magnetic bearing system. Specifically, signal analyzer's swept
sine function is applied experimentally to determine the transfer function of a single
input single output (SISO) path through the magnetic bearing system. Next, lead
compensator is designed in real time as to stabilize the operation of the system.

P. Barney et all. (2003) introduced an active control of a magnetically

levitated spindle. In this study, an unbalanced spindle was actively centered using an
Active Magnetic Bearing (AMB). To perform this task, modeling, simulation and
test program was implemented to design the Adaptive Least Mean Square (LMS)
controller. This study involves the implementation of LMS digital control algorithm
to maintain concentricity of an intentionally unbalanced spindle. The LMS
controller was implemented on an AMB system using a programmable digital signal
processor @SP). The LMS system model utilized the validated SIMULINK model
as a basis with the addition of an imbalance forcing function and LMS feed forward
algorithm. Finally, the LMS controller was implemented on the MBC 500
significantly improved the concentricity of the unbalanced shaft.

P. Rebecca and P. Gordon (2003) from Michigan Technological University

did a research based on disturbance rejection control of an electromagnetic bearing
spindle. Adaptive control algorithm is applied to MBC 500 magnetic bearing
system. Adaptive control is an appealing approach for the system because the
controller can tune itself to account for an unknown periodic disturbance, such as
cutting or grinding forces, injected into the system. An adaptive controller called the
Amplitude-Phase Adaptive Control Algorithm (APACA) was designed to augment
the lead-filter compensator. The purpose of APACA is to predict and compensate
for the external disturbance. This paper proved that an adaptive control algorithm
can be applied to an Active Magnetic Bearing (AMB) system with a periodic
disturbance applied to the rotor and resulting in minimal motion of the spindle. By
then, the position of the rotor can be stabilized.

It was followed by the research of Y. H. John (1995). He introduced a fuzzy

logic approaches to improve on dual acting magnetic bearing. The idea is to adjust the
linear controller signal in such a way that nonlinear effects are better compensated. The
relationships of attractive force to the electromagnet currents and air gap are described
and compensated using fuzzy principles. The fuzzy controller described in this section
was designed in two steps. First, fuzzy descriptions of the various operating points
which describe the antecedents and possible control adjustments as the consequents in
the fbzq control rules are computed. Second, a set of rules for control adjustments was
derived. The fbq rule outputs are composed using the max-min composition, and a
crisp value of an adjustment parameter was derived using the centroid method. The
design objective was to cancel the relationship of attractive force with respect to air gap
dimension. Finally, a fizzy controller which is able to adjust a winding current and
controlling a magnetic bearing system is designed successfully.
1.6 Layout Of Thesis

This section outlines the structure of the thesis.

Chapter 2 deals with the background of an active magnetic bearing system

which is the MBC 500. It includes the introduction, the advantages and
disadvantages besides the analysis and system modeling for the magnetic bearing
system. In addition, the mathematical modeling of the dynamic model of this
magnetic bearing is presented in details.

Chapter 3 discusses the state feedback control design theory which is applied
in the project such as the introduction of a state feedback controller, the
controllability and the observability of the system besides the technique in designing
a pole placement controller.

Chapter 4 explains the introduction of fuzzy logic structures such as the

fuzrtification, knowledge base and defuzzification besides the fuzzy control
approach.

Chapter 5 discusses the simulation results. The performance of the pole

placement controller and a fuzzy logic controller designed are verified and evaluated
using MATLAB-SIMULINK.

Chapter 6 concludes the topics and suggests recommendation for future

works.
 CHAPTER 2

2.1 Introduction MBC 500 Magnetic Bearing System

The MBC 500 is a desktop test bed made by Magnetic Moments

Incorporation and the system can be seen as shown as in Figure 2.1. It consist of a
horlzbntal stainless steel shaft or rotor which can be levitated using eight horseshoe
electromagnets, four at each end of the rotor [14].

There are two silver housings that hold the electromagnetic bearings which
levitate the spindle, so, the shaft nevet even touches the bearings when it is opeidting.
 Figure 2.1 :MBC 500 Magnetic Bearing System

Figure 2.2 shows the magnetic bearing. The copper-colored are the coils of wire
that make up the electromagnet. Each magnet has two "spokesn and is shaped like a
"U". There are four magnets in each bearing [14].

Figure 2.2 :Magnetic bearing

 +
--
. a,r--,ive
forces

I -+ spindle
I
- *
-
, -OYC --- -_ -I

Figure 2 3 :Attractive force exerted by electromagnet

One problem when using electromagnets is that they can only produce an
attractive force. The force is stronger when the spindle is closer to the magnet,
which in turn brings the spindle more close and makes the force even stronger. This
leads to an unstable system. As a result, the magnets must be arranged radially
around a spindle, so that a magnet on one side of the spindle can counteract the force
exerted by a magnet on the other side of the spindle. Moreover, a control system is
required to make the spindle levitate. The force exerted by the magnets can be
controlled by changing the current flowing through the coils of wire.

i2
fa-
g

where f = exerted force

i = current flowing through the coils of wire
g = distance between spindle and magnetic bearing

According to the Equation (2.1), a greater current increases the force exerted by the
magnet.
2.1.1 Advantages and Disadvantages of a Magnetic Bearing System

Magnetic bearing offer significant advantages because they do not come into
contact with other parts during operation, which can reduce maintenance. Higher
speeds, no friction, no lubrication, weight reduction, precise position control, and
active damping make them far superior to conventional contact bearings [8].

However, there are rephrase that limit the application of the magnetic bearing
such as to balance the electromagnets forces which are exerted on the magnetic
bearing and to maintain the position of the rotor at the equilibrium point [8]. Thus,
magnetic bearing need a controller which is able to stabilize the position of the rotor
during operation before it works successfblly.

2.1.2 Analysis and System Modeling for Magnetic Bearing System

The target system for this project to control is the Magnetic Moments MBC
500 Magnetic Bearing System. A diagram of this system is shown in Figure 2.4.
This system contains a stainless steel shaft or rotor which can be levitated using eight
horseshoe electromagnets, four at each end of the rotor. Hall E f f i t Sensors placed
just outside of the electromagnets at each end of the rotor to measure the rotor end
displacement. This system is a four degree of freedom system with two degrees of
freedom at each end of the rotor. These two degrees of freedom are translation in the
horizontal direction, perpendicular to the z axis (XI and x2) and translation in the
vertical direction (yl andyz) [14 ,231.
 An analysis of the geometry of the rotor will yield the following relationships:

XI = xo - (+-+in e (2.2)

x, = x o + (+-/)sin e

X , = xo - (4- 12)Sin e

X , = xo + ($-1,)sin B

Table 2-1 :System variables

Symbol Description
The displacement of center of mass of rotor
x, and xz The displacement of rotor at left and right bearings
XIand X2 The displacement of rotor at Hall Effect sensor
6 The angle that the long axis of the rotor makes with the z axis
FI and F2 The forces exerted on the rotor by left and right bearing

Table 2-2 :System parameters

Symbol Description Value
r
L Total length of the rotor 0.269m
1 Distance from each bearing to the end of the rotor 0.024m
Distance from each Halleffect sensor to the end of
12 0.0028m
the rotor
Moment of inertia of the rotor with respect to 1.5884 x 105
10
rotation about an axis in the y direction kg m2
m Mass of the rotor 0.2629kg
In general, moments and forces are interrelated in the following way

M=?XF

+
This relationship is shown pictorially in Figure 2.6(a). In this relationship r
+
is any vector pointing from the point to the line of application of the force F . If the
+ +
vector r is chosen to be perpendicular to the line of application of the force F as
shown in Figure 2.60>) then the above equation reduces to

1 i

Figure 2.6(a) Figure 2.6@)

Eigure 2.6 :Force / moment relation

Use the force and moment balance to derive the nonlinear differential
equations to describe the rigid body motion for the bearing rotor. It is assumed that
the motion is in only one plane (in the x direction only). The equations of motion are:
and the output equations can be expressed as

y, = X I = x , - 0.1317 sin 8

y, = X, = x , + 0.1317 sin 8

Moreover, linear equation, which described the dynamic system, is used in designing
a controller. The steps on linearizing the nonlinear equation are presented.

6 is assumed too small in order to linearize the equations of motion. Thus, the linear
motion equation of magnetic bearing system can be represented as;

where sin6 = Oand cos6 = 1.

A state-space system can be described as;

x = Ax+Bu

y=cx+Du
 CHAPTER 3

POLE PLACEMENT CONTROLLER DESIGN APPROACH

3.1 Introduction of Designing a Control System

Control system design is a multistage process involving. Before a controller

can be designed, the designer must have sufficient knowledge of the system to be
controlled. The designer begins by collecting information about the system from all
available sources and then representing this information in the form of a system
model. One source of information is the physics governing the system. An analysis
of these physical laws will yield differential or difference equations which describe
the motion of the system in response to certain input signals. Experimental input
butput data taken on the actual system is another common source of system
information. A given input signal is used to drive the systent arld the system
response to this signal is recorded. This experimental input output relationship then
is processed to determine a transfer fbnction model of the system [14].
 Once a suitable system model is obtained, the controller design can begin.
Typically, the initial design will be evaluated in computer simulation using a
software package such as MATLAB or SIMULINK. The performance and stability
of the controlled system are evaluated and design iterations of the controller are
performed until a controller is established which meets the design requirements [14].

This chapter discusses state-space design method based on the pole-

placement method. The pole-placement method is somewhat similar to the root-
locus method in that closed-loop poles are placed at desired locations. The basic
difference is that in root-locus design only the dominant closed-loop poles are placed
at the desired locations while in the pole-placement design, all closed loop poles are
placed at desired location [19].

3.2 State-space Repreedtation of ~ulti-Input-Multi-Output(MIMO) Systems

For linear multi-input-multi-output systems of order n with r inputs and m

outputs these equations become;

~ ( t=) A x(t) +B u(t) with the initial condition x(t,)

y(t) = C x(t) + D u ( t )

where the following notation is used;