Sie sind auf Seite 1von 16

MFE 3004 Mechatronics I Modelling Mechatronic Systems

MFE 3004 Mechatronics I

3. Modelling Mechatronic
Systems and Control

3.1 Introduction
The previous section has looked at design aspects and the relevant consideration of models that aid in
the various stages of design to both understand and analyse aspects of the design problem and its
solution. Indeed, a structured analysis model was adopted as a typical approach to analyse a complex
mechatronic system requirements. The use of structured analysis as a modelling tool can also be seen
within the context of designing the system information processing. However, structured analysis on its
own does not present an adequate representation of various system aspects, in particular behaviour
and dynamic performance characteristics of the physical part of the system. Alternative modelling
techniques may be additionally required for appropriate system analysis.

This section will look further into system modelling, with the aim of providing the basis for analysing
a system’s behaviour and its performance, mainly from a physical system perspective.

3.2 Goals of Modelling


In the previous section it was noted that models serve many purposes during the design of a
mechatronic system. Indeed, the structured analysis and the relevant requirements modelling that has
been adopted serves primarily to document and present in a clear fashion the requirements for which a
design solution is to be found. Models can also provide a hierarchical framework. Such a framework
ensures that the system as a whole is considered but in turn provides the basis of distinguishing the
various components of the system which will then allow the division of labour in the design process,
permitting concurrent work on separate components of the system. To a certain extent the role of the
structured analysis provides such a hierarchical framework through its dataflow diagram levels.
Furthermore, models can provide an appropriate insight into the behaviour of a system.

3.2.1 Documentation and Communication

Models present a simplification of the real system and its components and thus provide the basis for
summarising a lot of information. Thus, apart from being an effective tool in system analysis, they can
be seen as being the best mechanism for communication amongst all people involved in the design
and development of the system, from the design team members, to the customers and even to the
people involved in the manufacture of the designed system. Indeed, it is possible to identify four
distinct groups that stand to benefit from the utilisation of models as communication tools. The
engineering team needs communication amongst its members to ensure that effort is the design and
development is channelled to the elements and their interoperation that provide the main contribution
towards meeting the overall product requirements. In this respect one should include the
manufacturing personnel and their contribution towards ensuring a cost effective manufacture of the
designed product. Communication with the customer is also critical to ensure final customer
satisfaction in the designed and developed system. It is must that through appropriate communication,
the customer requirements are correctly understood and implemented in the design. Communication
with management is also fundamental in comprehending the status of the project , including the risks

C. Pace 3.1
MFE 3004 Mechatronics I Modelling Mechatronic Systems

undertaken, the costs as well as tradeoffs that would have to be made generally between performance,
functionality and cost. Finally, models can serve as a additional tool to store information on the design
and developments of products, thus serving as design process documentation which future design
teams can benefit from, particularly at earlier design concepts, analysis and decision stages.

When applying models, it should always be remembered that models are a simplification of reality,
utilised with a specific scope in mind, for example to understand system information and material
flow, or to provide a basis for analysing system physical dynamics, amongst other aspects. For this
purpose, certain models will be suitable for analysing certain system characteristics, which would be
clearly highlighted within the model, while other characteristics will be hidden and ignored within the
model. Therefore, if various system characteristics need to be analysed during the design process, it is
likely that various models will be used, each aiding in the analysis of specific distinct characteristics.
In this context, as a basis for communication, standardised modelling techniques as well as techniques
that are interdisciplinary (and not technology specific) ideally are applied wherever possible.

3.2.2 Hierarchical framework

Mechatronic systems in general include a large number of components that interact in many ways,
thus requiring various models as stated earlier on. The various models used would generally aim to
providing the means to subdivide complex problems into a set of simpler problems whose solutions
can potentially be integrated easily. This role of modelling would thus tend to suggest the use of
modelling techniques that allow the subdivision of the model into portions that can be independently
tackled by independent engineers. For example, systems of linear differential equations may provide a
substantial insight into the fundamental models of a system, but give little insight on how the design
problem should be divided among design team members. On the other hand, a model based on
dataflow diagrams can provide clear interfaces between elements that allow for system subdivision.

Additionally, when selecting models, consideration has to be additionally given to the mode by which
various models can be related together in order to allow cross-referencing between models. When
looking at modelling a system, one should therefore also take into consideration the possibility of
utilising compatible models or a unifying framework for the models, which will then provide a clearer
picture of the general system and its behaviour.

3.2.3 Providing Insights into system behaviour

Regarding insights into system behaviour, different models generally provide different types of
insight. For example differential equations provide insights regarding the time scales of various
behaviours and the relative importance of various factors as well as providing estimates of the validity
of simplifying assumptions. Other models can provide insight into the processing power that would be
required by the system. In each case the model provides insights into the problem or the
characteristics of a proposed solution by bringing into focus certain aspects of the modelled system
while hiding others from view.

3.3 Physical and Mathematical Modelling of


Dynamic Systems
On of the principle scopes for modelling is in the analysis of the system dynamics. A dynamic system
can be said to be any collection of interacting elements for which there are cause and effect
relationships among the time dependent variables. The analysis of the dynamic behaviour of physical
systems has become a keystone of modern technology and more than any other field links the
different engineering disciplines.

C. Pace 3.2
MFE 3004 Mechatronics I Modelling Mechatronic Systems

3.3.1 Stages of Dynamic Investigation

The purpose of any dynamic system analysis is to understand and predict the dynamic behaviour of a
given system and wherever possible to improve upon it. The following defines a list of the steps that
one generally goes through during the analysis of dynamic systems;

1. Physical Modelling – where the system to be studied is specified and a simple physical model is
adopted, the behaviour of which will match sufficiently closely the behaviour of the actual system.
2. Equations of Motion – here a mathematical model is derived to represent the physical model, i.e.,
the differential equations of motion of the physical model are appropriately derived.
3. Dynamic Behaviour –the dynamic behaviour of the mathematical model can be studied by solving
the differential equations of motion.
4. Design Decisions – Following the analysis design decisions can be made, i.e. the physical
parameters of the system that give the desired behaviour are chosen and implemented within the
design as well as the system is augmented, possibly with the introduction of appropriate controllers,
so that it will behave as desired.

The followings sections describe these characteristics in further detail

3.3.2 Physical Modelling : From the Actual System to the Physical Model

A physical model can be considered to be an imaginary physical system which resembles an actual
system in its salient features but which is simpler (more “ideal”) and is thereby more amenable to
analytical studies. When defining a physical model it is necessary to identify which simplifications
can be made, ensuring that the model still adequately reflects the actual system. In this respect,
engineering judgement plays a fundamental role. Knowing which actual system parameters are likely
to have a significant and insignificant effect on the system performance and applying appropriate
approximations that will not hinder the correct outcome of the analysis are indeed critical in develop
an appropriate physical model. Various approximations are used in physical modelling. A number of
these are given in table 3.1 below.

Table 3.1 Typical Approximations in Physical Modelling of Systems


Approximation Mathematical simplification

Neglect small effects Reduces number and complexity of differential


equations
Assume environment independent of system Reduces number and complexity of differential
motions equations
Replace distributed characteristics with Leads to ordinary (rather than partial) differential
appropriate lumped elements equations
Assume linear relationships Makes equations linear; allows superposition of
solutions
Assume constant parameters Leads to constant coefficients in differential
equations
Neglect uncertainty and noise Avoids statistical treatment

The table entries are described hereunder;

Neglect Small Effects - Small effects are neglected on a relative basis. In analysing the motion of an
airplane, we are unlikely to consider the effects of the earth’s magnetic field or gravity gradient. To
ignore these effects in a space vehicle problem gives grossly incorrect results.

Independent Environment – If we are modelling the influence of the environment on a system


component, in reality the system component itself may influence the environment, thus creating a

C. Pace 3.3
MFE 3004 Mechatronics I Modelling Mechatronic Systems

feedback loop. In modelling, such influences are commonly ignored to ensure simplicity of the model.
For example, in analysing the vibration of an instrument panel in a vehicle, we assume that the vehicle
motion is independent of the motion of the instrument panel.

Lumped Characteristics – In a lumped-parameter model, system dependent variables are assumed to


be constant over finite regions of space rather than over infinitesimal elements, as in a distributed-
parameter model. Note that elements in a lumped model do not necessarily correspond to separate
physical parts of the actual system. A long electrical transmission line has resistance, inductance, and
capacitance distributed continuously along its length. These distributed properties are approximated
by lumped elements at discrete points along the line.

Linear Relationships – Nearly all physical elements or systems are inherently nonlinear if there are no
restrictions at all placed on the allowable values of the inputs. If the values of the inputs are confined
to a sufficiently small range, the original nonlinear model of the system may often be replaced by a
linear model whose response closely approximates that of the nonlinear model. When a linear
equation has been solved once, the solution is general, holding for all magnitudes of motion. Solutions
to linear equations may be superposed on one another.

Constant Parameters – Time-varying systems are ones whose characteristics change with time.
Physical problems are simplified by the adoption of a model in which all the physical parameters are
constant over time.

Neglect uncertainty and noise – In real systems we are uncertain, in varying degrees, about values of
parameters, about measurements, and about expected inputs and disturbances. Disturbances contain
random inputs, called noise, which can influence system behaviour. It is common to neglect such
uncertainties and noise and proceed as if all quantities have definite values that are known.

The most realistic physical model of a dynamic system leads to equations of motion that are;
nonlinear, partial differential equations with distributed parameters, with time-varying and space-
varying parameters and which include a representation of the uncertain parameters of the system.
These equations however are the most difficult to solve and in most practical cases only numerical
analysis of the equations is possible. The simplifying assumptions just discussed lead to a physical
model of a dynamic system that is less realistic and to equations of motion that are; linear, ordinary
differential equations with constant coefficients and without uncertainty factors represented. This type
of equation is much easier to solve and the most general and highly developed theory for control
design is based on this type of differential equation. However if such models tend to deviate from the
real situation, and thus give results of system dynamic behaviour that deviate from the real physical
system to be designed, then why are such models developed in the first place? The key in this case
lies with engineering judgement, i.e. knowing when the model is sufficiently accurate to provide an
adequate analysis of the system under investigation. In a number of instances linear ordinary
differential equations may suffice in providing an adequate system representation, in other instances it
may be necessary to opt for alternative more complex models. It is up to the design engineer to
develop a model that can adequately represent reality to an extent that will allow the engineer to
derive appropriate design decisions.

In section 3.4 these modelling aspect choices will be discussed further.

3.3.3 Equations of Motion : From Physical Model to Mathematical Model

A natural extension of the physical model for analytical purposes is the derived mathematical model,
for which a great deal has already been said during the previous discussion on physical models. The
focal points in deriving the equations of motion that define the mathematical model for a given
physical system model can be stated as follows;

C. Pace 3.4
MFE 3004 Mechatronics I Modelling Mechatronic Systems

Dynamic Equilibrium Relations – writing of dynamic equilibrium relations to describe the balance of
forces, or flow rates, of energy, etc., which must exist for the system and its subsystems. System and
subsystem differential equations describe such dynamic equilibria.

Compatibility Relations – writing of system compatibility relations to describe how motions of the
system elements are interrelated because of the way they are interconnected. These are inter-element
or system relations. Thus transmission ratio and or related motion converter relations relate to such
compatibility relations as well as relations related to interconnected electrical and electronic
components.

Two further considerations are involved in deriving equations of motion, these being

Physical variables – selection of precise physical variables (velocity, voltage, pressure, flowrate etc.)
with which to describe the instantaneous state of a system, and in terms of which to study its
behaviour.

Physical Laws – natural physical laws which the individual elements of the system obey, including;
• Mechanical relations between force and motion
• Electrical relations between current and voltage
• Electromechanical relations between force and magnetic field
• Thermodynamic relations between temperature, pressure, internal energy, etc..

These relations are called constitutive physical relations as they concern only individual elements or
constituents of the system.

3.3.4 Summary of Mathematical Modelling Building Blocks

Table 3.2 describes the basic building blocks for modelling various systems, including mechanical,
electrical, fluid and thermal system elements.

Table 3.2 Mathematical modelling building blocks for Physical modelling elements
Physical Element/ Modelling Constants/ Describing Equation Energy Stored (E)/
Building Block Variables Power Dissipated
(P)
Mechanical Elements
Translational – F =- force F = kx 1 F2
Spring k = spring constant E=
x = displacement 2 k
Translational - F = Force dx P = bv 2
Dashpot b = damping constant F =b
x = displacement dt
v = velocity
Translational – F = Force d 2x E = 12 mv 2
Mass m = mass F =m
x = displacement dt 2
v = velocity
Rotational - Spring T = torque T = kθ 1 T2
k = torsional spring const. E=
θ = angular displacement 2 k
Rotational - T = torque dθ P = bω 2
Damper b = torsional damping const. T =b
θ = angular displacement dt
ω = angular velocity
Rotational –Inertia T = torque d 2θ E = 12 Iω 2
I = moment of inertia T =I
θ = angular displacement dt 2
ω = angular velocity

C. Pace 3.5
MFE 3004 Mechatronics I Modelling Mechatronic Systems

Electrical Elements
Resistance i = current V V2
V = voltage i= P=
R = resistance R R
Inductance i = current 1 E = 2 Li 2
1

L∫
V = voltage i= Vdt
L = inductance
Capacitance i = current dV E = 12 CV 2
V = voltage i=C
C = capacitance dt
Fluid Power Elements
Hydraulic q = flow rate Δp 1 2
Resistance Δp = pressure difference q= P= Δp
(Resistance to flow) R = flow resistance* R R
Hydraulic Fluid q = flow rate d Δp E = 12 C Δp 2
Capacitance Δp = pressure difference q=C
(Fluid storage) dt
A
C = fluid capacitance =
ρg
Hydraulic Fluid q = flow rate 1 E = 12 Iq 2
I∫
Inertance Δp = pressure difference q= Δpdt

I = fluid inertance =
A
Pneumatic m& = mass flow rate Δp 1 2
Resistance Δp = pressure difference m& = P= Δp
R R
R = mass flow resistance
Pneumatic m& = mass flow rate dp E = 12 Cp 2
Capacitance p = pressure m& = C
dt
C = pneumatic capacitance =
V
(constant volume)†
RT
Pneumatic Fluid m& = mass flow rate 1 E = 12 Im
&2
I∫
Inertance Δp = pressure difference m& = Δpdt
L
I = fluid inertance =
A
Thermal Elements
Thermal q = heat flow rate dT E = CT
Capacitance T = temperature q=C
C = thermal capacitance =Mc dt
Thermal Resistance q = heat flow rate ΔT
ΔT = temperature difference q=
R = thermal resistance R

3.3.5 General Mathematical Standard Models of Systems

As stipulated earlier on, it is quite common to model systems using standard, ordinary linear
differential equations. This is so since a large amount of analysis and a substantial insight into
controller development for such system models has been performed on these types of mathematical
models. The standard linear ordinary differential equation can be represented as follows;

d n qo d n −1qo dqo d m qi d m −1qi d 1qi


an + an −1 + ... + a + a q = b + bm −1 + ... + b + b0 qi
dt n −1 dt m−1
1 0 o m 1
dt n dt dt m dt1

*
A linear flow resistance relationship is assumed whereas in reality non-linear relationships are needed to model
turbulent flow and flow across sharp-edged orifices.

The equation is only valid for a constant volume container.
C. Pace 3.6
MFE 3004 Mechatronics I Modelling Mechatronic Systems

In the above equation;


qo is the output (response) variable of the physical system
qi is the input (excitation) variable of the physical system
an, and bm, etc.. are the physical parameters of the system and are considered as constants.

In general it is quite common to try to model systems as either zero, first or second order systems. The
equivalent equations for such models are as follows;

Zero-order dynamic system model; a0 qo = b0 qi


dq dq
First-order dynamic system model; a1 o + a0 qo = b1 i + b0 qi
dt dt
2
d q dq d 2q dq
Second-order dynamic system model; a2 2o + a1 o + a0 qo = b2 2 i + b1 i + b0 qi
dt dt dt dt

In general, such system models dynamic performance is evaluated through the analysis resulting from
specific system inputs qi. Typical examples in this case would include a step input, ramp input,
impulse input as well as a sinusoidal input. Figure 3.1 illustrates the typical dynamic response curves
for a zero order, first-order and second-order system when the input is a step response (i.e. qi =
constant).
qo(t) qo(t)

K K

qo(t)=K

K(1 - e τ)
-t/

Time t Time t
(a) (b)
Output qo(t)
steady state output

2.0
ξ = 0.0
ξ = 0.1
ξ = 0.2
1.8

ξ = 0.3
1.6
ξ = 0.4
ξ = 0.5
1.4
ξ = 0.6
ξ = 0.7
1.2

1.0

0.8
ξ = 0.8
0.6
ξ = 0.9
ξ = 1.0
0.4
ξ = 2.0
0.2

0
0 1 2 3 4 5 6 7 8 9 10 11 12 ωnt
(c)

Figure 3.1 System step input response for (a) zero-order system, (b) 1st-order system, (c) 2nd-order system.

C. Pace 3.7
MFE 3004 Mechatronics I Modelling Mechatronic Systems

For the case where the input varies in a sinusoidal manner, system analysis is generally performed
over a range of input signal frequencies qi = q iO sin (ω t ) . The resulting output for a sinusoidal input
qi = q iO sin (ω t ) , will also be a sinusoidal signal, with the same frequency but with a varying
amplitude and phase shift, qo = q oO sin (ω t + φ ) . Thus, the amplitude ratio qiO / q oO and phase shift
φ are generally evaluated with respect to the change in frequency of the input (i.e. the frequency
response).

3.3.6 Example of physical and mathematical modelling: Electrodynamic vibration


shaker

Figure 3.2 illustrates an electrodynamic vibration shaker. This ‘moving coil’ type of device converts
an electrical signal into a mechanical force and/or motion and is very commonly used in vibration
shakers such as vibratory bowl feeders. Here the current-carrying coil is located in a permanent
magnetic field which is either generated by a permanent magnet in small devices or by an electrically
excited coil in larger applications.
Table
Threaded Insert
Drive Coil
Rubber Dirt
Shield

Tangential
Flexure

Radial
Flexure

Magnetic
Structure Permanent
Magnet

Rubber
Feet

Figure 3.2 Schematic of an electrodynamic vibration shaker

Two electromechanical effects are observed in such configurations:


• The motor effect which is the motion of the coil through the magnetic field, causing a voltage
proportional to velocity to be induced into the coil.
• The generator effect which is the passage of current through the coil, causing it to experience
a magnetic force proportional to the current.
Full-scale force capabilities for different models range from about 20 to 150,000 N with permanent
magnet fields being used below about 250N, Figure 3.3 illustrates the physical model applied to the
electrodynamic device.

From figure 3.3, flexure Kf is an intentional soft spring (yet stiff in the radial direction) that serves to
guide the axial motion of coil and table. Flexure damping Bf is usually intentional, fairly strong, and
obtained by laminated construction of the flexure spring, using layers of metal, elastomer, plastic, and
so on. The coupling of the coil to the shaker table would be rigid so that magnetic force is transmitted
undistorted to the mechanical load. Thus Ktc (generally large) and Btc (quite small) represent unwanted
effects rather than intentional spring and damper elements. R and L are the total circuit resistance and
inductance, including contributions from both the shaker coil and the amplifier output circuit.
C. Pace 3.8
MFE 3004 Mechatronics I Modelling Mechatronic Systems

Mt Table xt

Coil/Table
coupling
Flexure Kf Ktc Flexure
Btc Bf
Stiffness Damping

S N S xc
Mc

Coil Magnet

Foundation

i
+
Coil/ power amplifier
ei parameters

R L
Figure 3.3 Physical Model of the electrodynamic vibration shaker

The mathematical model shown hereunder is obtained by application of Newton’s law to the table and
coil masses and by application of Kirchoff’s voltage law to the electrical circuit. The input is the
output voltage ei, of the electronic power amplifier that drives the shaker, and the outputs are the
displacements of the table and coil, xt and xc, respectively, and the electrical circuit current, i.

− K f xt − B f x&t − K tc ( xt − xc ) − Btc ( x&t − x&c ) = M t &&


xt
K tc ( xt − xc ) + Btc ( x&t − x&c ) + K f / i i = M c &&
xc
di
−ei + iR + L + K e / x& x&c = 0
dt

K e / x& = K f / i represents the electro-mechanical coupling of the coil in V/(m/s) or in N/Amp such that;

Coil back e.m.f. = K e / x& x&coil


Or Coil force = K f / i icoil

From these equations the mathematical model for the system is defined. Through solving such
equations, the frequency response of the system can be appropriately evaluated for typical values of
the various constant parameters. Thus for various values of the voltage ei magnitude and frequency,
the corresponding vibration oscillations and frequency can be determined by viewing the mode in
which the displacement values, xt in particular, will vary.

C. Pace 3.9
MFE 3004 Mechatronics I Modelling Mechatronic Systems

3.4 A Summary of choices for System


Modelling
As has been highlighted throughout the previous discussion, there are a number of decisions that must
be made in selecting a model for the behaviour of a dynamic system. These include the choice
between analytical and numerical modelling, distributed or lumped parameters, the approach used to
model random factors as well as the choice between linear and non-linear models. These are
summarised hereunder.

3.4.1 Analytical vs Numerical Models

An initial distinction in the type of model to be adopted can be made between the use of analytical or
numerical models. Analytical models are models are generally based on some mathematical
formulation of the system characteristics, these generally taking the form of a set of linear differential
equations when modelling system dynamics. Such models provide insights into overall system
behaviours just from the mathematical formulations themselves. An advantage of such models is that
they allow a comparison to be made between systems that are dissimilar but which have similar
models. A typical example can be seen between mechanical and electrical models, the performance of
both being potentially modelled through an appropriate differential equation. These analogous system
models often prove useful in the early design stage of concept generation since they provide the
means to highlight potential solutions to a problem, which may originate from a completely different
system, but which exhibits similar model properties. An additional advantage of such models is that
the characteristics of a whole set of systems with the same analytical model can be analysed and
solutions defined for the whole set. Indeed, substantial data and information on system models based
on linear differential equations is readily available. However a major problem with such models,
particularly when having to model complex systems, is that the analytical model are generally
substantial simplifications that ignore non-linearities and time-dependent behaviour, factors that may
greatly affect system overall behaviour. Therefore when applying analytical models, one must be
careful to explicitly state assumptions made in the modelling process and ensure that such
assumptions do not distort the model to the extent that the model does not provide an adequate
reflection of reality.

Numerical analysis, on the other hand is limited in the insight it offers on the ‘internal’ mechanisms of
the system. However it provides more detailed insights into real system behaviour and performance. It
is quite common that such numerical analysis tools are applied at a second stage following the
application of more analytical approaches. For example, a non-linear system might initially be
analysed using a linearised model with appropriate optimal solutions. These solutions can then be
evaluated through the application of numerical tools, directly on the non-linear model, and the
solution is appropriately adapted to suite the non-linearities. Thus, the availability of such numerical
methods, particularly through the use of appropriate computer simulations, opens many doors to
modelling options that would not otherwise be considered due to their complexity, providing an
additional source in design improvement. For example, numerical solutions for nonlinear differential
equations allow modelling of systems that had previously been approached only with linear
approximations.

3.4.2 Partial vs Ordinary Differential Equations

When modelling system dynamics, it is quite often necessary to determine whether the system can be
adequately modelled through ordinary differential equations, or whether partial differential equations
need to be used for appropriate system modelling. Partial differential equations are used whenever the
variables or parameters of interest in the analysis will vary spatially as well as temporally (i.e. they
vary with respect to another parameter apart from time). If, for example, a robotic arm is considered,
and it is assumed that no deflection occurs under load within the robotic arm members (i.e. the arm
members are rigid), then ordinary differential equations can be used to be able to model the dynamics
of the arm. However, if arm member bending under load is considered (which will undoubtedly be a

C. Pace 3.10
MFE 3004 Mechatronics I Modelling Mechatronic Systems

truer reflection of reality), then partial differential equations will have to be used. However, if the arm
flexibility needs only to be considered in terms of the end effector position, it is possible to use a
lumped parameter model for which ordinary differential equations can then be applied. The lumped
parameter model can be used to summarise the arm dynamics in terms of end effector motion and
dynamic forces acting on the end effector. Yet, there should be no interest by the designer in the
detailed variation of the parameter over space. If such a need for detailed analysis is required, such as
for example, the stress concentration or the temperature distribution, the simplified lump sum model
cannot be applied.

3.4.3 Stochastic vs Deterministic Models

Another issue that has to be taken into consideration in dynamic system modelling is that related to
uncontrolled variability within the system or as a disturbance on the system. Uncontrolled variability
can arise due to numerous factors. Knowing and understanding the characteristics of such variability
will provide a significant step in comprehending how such variability should be modelled. If the
variability is known is exact terms, then it is possible to model such variability in a deterministic
manner, without the need to model any uncertainty into the parameter variation. On the other hand,
certain parameter characteristics may not be known with certainty and it would be necessary to
consider such uncertainty within the model adopted. For example, some system parameters may not
be known with an adequate precision but are known to remain constant over time. In this case an
analysis can be performed for various parameter values in order to determine the sensitivity of the
system to such parameter value changes. In other cases, it might be known that the parameter will
vary slowly with time, although the mode of such variation in terms of speed and magnitude of
change may not be known. In this case, one must examine the dynamic nature of such parameter
variation in order to determine if the system can be analysed in quasi-steady state or if the dynamics
of the parameter variations are such that they have to be coupled with the dynamics of the system.

When it is necessary to consider such uncertain parameter variations over time, appropriate parameter
variation models may be required. These can either be in the form of an arbitrary variable parameter
input value, or else if certain statistical properties of the parameter value are known, then these
properties can be modelled in the parameter variation. In these cases, stochastic models are generally
adopted, where the parameter variation uncertainty is explicitly represented in some manner.

3.4.4 Linear vs Non-linear

Linear and non-linear models have already been mentioned earlier on. Indeed, the choice between
applying a linear model and a non-linear model for system dynamic performance analysis will
generally be a major decision in terms of both model development as well as the analytical approach
to be adopted. Linear models tend to be simpler models where a substantial number of techniques are
available for the analysis of such linear time invariant systems. However, only very few simple
engineering systems are in reality linear, and as stated earlier, the application of linear models to non-
linear systems may not prove adequate enough for system analysis and design. One way to overcome
such problems is by dividing the complex system into simpler sub-components, allowing the potential
isolation of non-linearities. This is shown in figure 3.4(a). Additionally, for small variations about a
differentiable point on a non-linear curve, one can always find a linear model, but the value of this
model might be extremely limited depending on its accuracy over a reasonable range of input values.
This is shown in figure 3.4(b) with a linearised model is taken at a specific operating point.

When dealing with a discontinuous or non-differentiable non-linearity, however, a greater problem


arises. A linear model can potentially be found only for the system when it is operating far from the
discontinuity region or when the discontinuity is small compared to the linear portion of the model.
However, on approaching the discontinuity linear behaviour cannot be assumed any more.

C. Pace 3.11
MFE 3004 Mechatronics I Modelling Mechatronic Systems

Linear operating
region Linearised
relationship with
slope defined at
point P

Non-Linear Non-Linear
region region

Operating
Point P

(a) (b)

Figure 3.4 Dealing with non-linearities (a) Isolating the non-linearities by considering only the linear portion of
a relationship. (b) Assuming a linear relationship about an operating point by taking the relationship to be
represented by the slope at the specified point.

Non-linearities generally also present and model system behaviours that are not possible in purely
linear systems. Non-linear models are thus needed in these circumstances.

3.5 The Role of Control in Mechatronics


Mechatronics systems exhibit some form of embedded control system. The embedded control
generally takes the form of programmable electronics, such as embedded microcontrollers and
computers. The distinction of mechatronic systems from other controlled mechanical systems tends to
occur in the fact that whereas in other systems, the controller seems to be an add-on to the mechanical
system, generally following the design of the mechanical system, in mechatronics controller design is
done simultaneously with all other design features of the whole system. Thus, from a purely control
engineering perspective, the only variable(s) from the design perspective is the controller
parameter(s), generally based on some form of optimisation function which is based on a set of
constraints defined by the pre-designed process to be controlled. This is illustrated in figure 3.5 (a)
below.

On the other hand, in mechatronic system design, the whole optimisation process is not linked solely
to the controller parameters, but rather to the system as a whole. Hence, on designing and evaluating
control mechanisms, the mechatronics engineer will consider the process, its measurement and its
control as all being variables to be optimised based solely on the constraints defined by the original
customer requirements. This is shown pictorially in figure 3.5(b).
Optimisation
+ +
Controller Process Controller Process
_ _
Optimisation

Interface Sensor(s) Interface Sensor(s)

(a) (b)

Figure 3.5 (a) Optimisation of the controller in control engineering, (b) optimisation of all system components
within a mechatronics approach.

C. Pace 3.12
MFE 3004 Mechatronics I Modelling Mechatronic Systems

In general, the optimisation of the process as a whole does not occur simultaneously. As discussed
when dealing with modelling techniques, complex mechatronic problems generally need to be
subdivided into smaller problems, with appropriate interface definitions, in order to achieve an
appropriate solution. However, the continuous evolution of the sub-problems and their solutions, and
the integration of such solutions should target the optimisation of the system as a whole rather than of
its individual components, including the controller, and therefore controller design will have to go
hand in hand with the design of the rest of the system.

3.5.1 Types of Controllers in Mechatronic Systems

There are various types of controllers that can be used in integrated systems. Classical controllers are
generally classified broadly into open-loop and closed loop controllers, where in general closed-loop
control tends to give much better system performance characteristics than open-loop systems.
Additional controller parameters are sometimes included to enhance controller performance, such as
the use of feed-forward control loops, generally used for measuring system disturbances and
modifying controller behaviour according to such system disturbances.

Figure 3.6 reiterates the characteristics of a mechatronic system in block form, depicting the control
loops within the mechatronic system. The role of the embedded computer based controller in such
systems is always to enhance the mechanical behaviour of the system. This generally also entails the
presence of ‘intelligent’ characteristics in the system, mainly through advanced forms of control

Feedforward
Other knowledge and (disturbance
information sources compensator)

+
Setpoint + Feedback +
generator Controller DAC Actuator Process
_
Sensor
Fusion ADC Sensor

Information Domain Energy Domain

Figure 3.6 Basic layout of a mechatronic control system

Further advancements in controllers allow the controllers to compensate for system non-linearities,
such as through the use of adaptive control as well as learning forms of control.

3.5.2 Defining Controlled System Performance Specifications

There are various requirements that need to be defined when coming to develop a controlled system.
Performance specifications are definitely amongst the most important. However, as always, due
consideration needs to be given to practical economic criteria when coming to design and develop a
system. Thus, although the focus throughout this chapter and this section has been and is on the
performance characteristics, it is impossible to ignore cost issues and the engineering designer has to
be cognizant of these higher-level considerations that will eventually influence the system design,
potentially resulting in a compromise on system performance.

With regards to performance requirements for controlled systems, it is possible to visualise the
controlled system objective as being one where the system output follows a desired value, whilst
eliminating the effect of any disturbances. Therefore it is logical to expect controlled system
performance criteria to measure how well the system output follows the desired value as well as its
ability to eliminate disturbance effects.

C. Pace 3.13
MFE 3004 Mechatronics I Modelling Mechatronic Systems

Unfortunately, it is highly unlikely that exact system desired inputs and disturbances are known in
advance, including how such values will vary in time. Ideas of the range of values may be available as
well as some details on their dynamic characteristics but it is practically very difficult to evaluate the
designed controlled system behaviour from actual values of the input signals and disturbances
experienced by a real physical system. As stipulated earlier on when discussing system models, it is
therefore common to base the performance evaluation of system responses on a set of simple
‘standard’ inputs such as the step, ramp and sine wave input. The use of standard signals has been
quite successful for several reasons, including;
- there is a good correlation for various classes of control systems between the response of the
system to standard inputs and the ability of the system to accomplish practical tasks
- the design process revolves around the need of comparing various competitive designs. The
use of standard signals and performance measurements corresponding to these standard
signals tends to make this comparison process much easier.
- The use of standard inputs tends to make the eventual mathematical analysis much more
simple and straight forward, and therefore more amenable to verification

The general performance criteria are thus based on these standard input test signals and are generally
grouped under;
Time-domain performance criteria – criteria concerned with the response to step inputs, ramp inputs
and the like, and
Frequency-domain performance criteria – criteria concerned with certain characteristics of system
frequency response.

Both the time-domain and frequency-domain design criteria generally are intended to specify one or
the other of;
• Speed of response
• Relative system stability to disturbances
• Steady-state errors

Amongst the time-domain performance criteria, the following can be specified in terms of a step
input;
Rise Time Tr – the time for the system output to go from 10% to 89% of the desired value
Peak Time Tp – the time for the controlled system output to reach its highest value
Percentage Overshoot Op (sometimes given as the maximum overshoot) – the maximum system
output reached beyond the desired value.
Settling Time Ts – a measure of the time taken for the system response to settle, generally to within
5% of the desired value.
Steady State Error Ess – the error in the system output when compared to the system desired value one
the system has settled down and ‘entered the steady-state region’.

The rise time and peak time are generally used for measuring the speed of response, i.e. how quickly
the system starts responding to a change in the desired value. The overshoot value generally gives a
reflection of system stability (and is generally kept to within 10% to 20%). The settling time reflects
both system stability as well as response speed. The steady-state error on the other hand reflects the
controller’s ability to track the desired value whilst in the steady state. Figure 3.7 illustrates these
measures.

Defining such measures is important in allowing an appropriate definition of the controlled system
behaviour. For example, if figure 3.8 is considered, it is quite difficult to determine when system A or
B is the faster. In terms of rise time A would be defined as the faster system, whereas in terms of
settling time, B would be defined as the faster system response. Clear criteria must thus be stipulated
for determining the optimal system response for the controlled system being designed.

C. Pace 3.14
MFE 3004 Mechatronics I Modelling Mechatronic Systems

Figure 3.7 Time-domain response criteria for a closed-loop system response. Curve C represents the system
output whereas V represents the system input. A zero disturbance is assumed.

Figure 3.8 An ambiguous definition of a system’s response speed. Is system A or system B the faster system to
respond to the step input?

Similar criteria can be set from the frequency response perspective, including the system bandwidth
which is a measure of the frequency range in which the system responds adequately without
substantial response deterioration.

3.6 Summary
System modelling is an integral part of mechatronic system design since it allows the achievement of
various objectives. Initially, modelling allows for an appropriate system representation that avoids
ambiguities and can serve as a tool for appropriate communication amongst design team members as
well as with the customer. Furthermore, system models provide the tools by which the designed
system can be appropriately analysed. Modelling tools that allow the depiction of various
technologies tend to be very useful from a mechatronic perspective, where technology integration is a
must. Amongst the various models used, mathematical system models, in their various forms, tend to
be the most amenable for analysis. When coming to analyse system dynamics, these mathematical
models generally take the form of differential equations. Various considerations need to be taken
when developing the appropriate mathematical models, ensuring that the model adequately reflects
the real system it is representing, but at the same time, is simple enough to allow appropriate analysis.
Developing such models with the relevant assumptions thus becomes a fundamental role of the design
engineer and analyst. Finally, when considering control in mechatronic systems, it is necessary to

C. Pace 3.15
MFE 3004 Mechatronics I Modelling Mechatronic Systems

view controller development as an integral part of the whole system development. Appropriate
controlled system performance criteria must be defined, against which the system performance can be
measured and from which it is possible to determine the adequacy of the modelled controlled system
and eventually of the physical system to be implemented.

References

Mechatronics Handbook , Robert H. Bishop (Ed.) , CRC Press, 2002


Mechatronics Systems Design with Applications, Course book, Kevin C. Craig, Fred P. Stolfi, the
American Society of Mechanical Engineers (ASME), the Institute of Electrical and Electronic
Engineers, USA (IEEE), 1993
Mechatronics, Dan Necsulescu, Prentice Hall, 2002

C. Pace 3.16

Das könnte Ihnen auch gefallen