Sei sulla pagina 1di 17

MIMO Control

(Multiple Input Multiple


Output)
Introduction to MIMO Control system, history, applications and the Case
Study of MIMO Control.

Mohsin Ali
2011-MS-CH-68

Sohail Aziz
2011-MS-CH-65
University of Engineering &Technology
Lahore
/2012

Introduction
Many complex engineering systems are equipped with several actuators that
may influence their static and dynamic behavior. Commonly, in cases where
some form of automatic control is required over the system, also several
sensors are available to provide measurement information about important
system variables that may be used for feedback control purposes. Systems
with more than one actuating control input and more than one sensor output
may be considered as multivariable systems or multi-input-multi-output
(MIMO). The control objective for multivariable systems is to obtain a
desirable behavior of several output variables by simultaneously
manipulating several input channels.
Systems with more than one input and/or more than one output are known
as Multi-Input Multi-Output systems, or they are frequently known by the
abbreviation MIMO. This is in contrast to systems that have only a single
input and a single output (SISO).

Design Question for MIMO Control

Consider a general process with several inputs and outputs (Figure). Several
questions must be answered before we design a control system for such a
process.

1- What are the control objectives? In other words, how many and which of all
possible variables should be controlled at desired value? This
seemingly simple question is critical for the design of efficient control
system.
2- What outputs should be measured? Once the control objectives have been
identified we need to select the measurements necessary to monitor
the operation of the process. We can classify the measured outputs
into two categories:
a) Primary measurements: These are the controlled outputs through
which we can determine directly if the control objectives are
satisfied.
b) Secondary measurements: These are not used to monitor directly
the control objectives but are auxiliary measurements employed
for the cascade, adaptive, or inferential control.
3- What inputs can be measured? We assume that all the manipulated
variables are measurable and there for can be employed for adaptive
model (model reference for the self-tuning regulator) and inferential
control. With respect to the disturbances only a few can be measured
easily, rapidly and reliably. These measurable disturbances can be
used to construct feed-forward, feedforward-feedbackward and ratio
control configurations.
4- What manipulated variables should be used? A multiple-input, multiple-output
system possesses several manipulated variables which can be used to
design of the control system. The selection of most appropriate
manipulations is very critical problem and should be approached with
care. Some manipulations have the direct, fast and strong effect on the
controlled outputs; others do not. Furthermore, some variables are
easy to manipulate in real life(e.g, liquid flow); others are not(e.g. flow
of solids, slurries, etc)
5- What is the configuration of the control loop? Once all the possible
measurements ad manipulations have been identified, we need to
decide how they are going to be interconnected through the control
loops. In other words, what measurement will accurate a a given
manipulated variable or what manipulation will be used to regulate a
given controlled outputs at its desired value?

Example of Multivariable Feedback Systems

The following two examples discuss various phenomena that specifically


occur in MIMO feedback systems and not in SISO systems, such as
interaction between loops and multivariable non-minimum phase behavior.

Figure: Two-tank liquid flow process with recycle


Figure: Two-loop feedback control of the two-tank liquid flow process
Consider the flow process of Fig. The incoming flow 1 and recycle flow 4
act as manipulable input variables to the system, and the outgoing flow 3
acts as a disturbance. The control objective is to keep the liquid levels h1 and
h2 between acceptable limits by applying feedback from measurements of
these liquid levels, while accommodating variations in the output flow 3. A
dynamic model, liberalized about a given steady state, is

Here, hi and i denote the deviations from the steady state levels and flows.
Laplace transformation yields

which results in the transfer matrix relationship

The example demonstrates the following typical MIMO system phenomena:

Each of the manipulable inputs 1 and 4 affects each of the outputs


to be controlled h1 and h2, as a consequence the transfer matrix

Consequently, if we control the two output variables h1 and h2 using


two separate control loops, it is not immediately clear which input to
use to control h1, and which one to control h2. This issue is called the
input/output pairing problem.

A possible feedback scheme using two separate loops is shown in


Figure. Note that in this control scheme, there exists a coupling
between both loops due to the non-diagonal terms in the transfer
matrix. As a result, the plant transfer function in the open upper loop
from 1 to h1, with the lower loop closed with proportional gain k2,
depends on k2,
Here llc indicates that the lower loop is closed. Owing to
the negative steady state gain in the lower loop, negative values of k2 stabilize the lower loop. The dynamics of P11.
(s)/ llc changes with k2

The phenomenon that the loop gain in one loop also depends on the loop
gain in another loop is called interaction. The multiple loop control structure
used here does not explicitly acknowledge interaction phenomena. A control
structure using individual loops, such as multiple loop control, is an example
of a decentralized control structure, as opposed to a centralized control
structure where all measured information is available for feedback in all
feedback channels.
In the example the input-output pairing has been the natural one: output i is
connected by a feedback loop to input i. This is however quite an arbitrary
choice, as it is the result of our model formulation that determines which
inputs and which outputs are ordered as one, two and so on. Thus the
selection of the most useful input-output pairs is a non-trivial issue in
multiloop control or decentralized control, i.e. in control configurations where
one has individual loops as in the example. A classical approach towards
dealing with multivariable systems is to bring a multivariable system to a
structure that is a collection of one input, one output control problems. This
approach of decoupling control may have some advantages in certain
practical situations, and was thought to lead to a simpler design approach.
However, as the above example showed, decouplingmay introduce additional
restrictions regarding the feedback properties of the system.

Survey of developments in multivariable control

The first papers on multivariable control systems appeared in the fifties and
considered aspects of noninteracting control. In the sixties, the work of
Rosenbrock (1970) considered matrix techniquesto study questions of
rational and polynomial representation of multivariable systems.
Thepolynomial representation was also studied by Wolovich (1974). The
books by Kailath (1980)and Vardulakis (1991) provide a broad overview. The
use of Nyquist techniques for multivariablecontrol design was developed by
Rosenbrock (1974a). The generalization of the Nyquistcriterion and of root
locus techniques to the multivariable case can be found in the work
ofPostlethwaite and MacFarlane (1979). The geometric approach to
multivariable state-space controldesign is contained in the classical book by
Wonham (1979) and in the book by Basile andMarro (1992). A survey of
classical design methods for multivariable control systems can befound in
(Korn and Wilfert 1982), (Lunze 1988) and in the two books by Tolle (1983),
(1985).Modern approaches to frequency domain methods can be found in
(Raisch 1993), (Maciejowski1989), and (Skogestad and Postlethwaite 1995).
Interaction phenomena in multivariable processcontrol systems are
discussed in terms of a process control formulation in (McAvoy 1983).
Amodern, process-control oriented approach to multivariable control is
presented in (Morari andZafiriou 1989). The numerical properties of several
computational algorithms relevant to the areaof multivariable control design
are discussed in (Svaricek 1995).
A CaseStudy for Helicopter Control

In this section, we present a case study which illustrate a number of


important practicalissues, namely: weights selection in Hmixed-sensitivity
design, disturbance rejection,output selection, two degrees-of-freedom H
loop-shaping design, ill-conditioned plants, analysis and synthesis.

The complete design process for an industrial control system will normally
includethe following steps:

1. Plant modelling: to determine a mathematical model of the plant either


from experimental data using identification techniques, or from
physical equations describing the plant dynamics, or a combination of
these.
2. Plant input-output controllability analysis: to discover what closed-loop
performance can be expected and what inherent limitations there are
to good control, and to assist in deciding upon an initial control
structure and may be an initial selection of performance weights.
3. Control structure design: to decide on which variables to be
manipulated and measured and which links should be made between
them.
4. Controller design: to formulate a mathematical design problem which
captures the engineering design problem and to synthesize a
corresponding controller.
5. Control system analysis: to assess the control system by analysis and
simulation against the performance specifications or the designers
expectations.
6. Controller implementation: to implement the controller, almost
certainly in software for computer control, taking care to address
important issues such as anti-windup and bumpless transfer.
7. 7. Control system commissioning: to bring the controller on-line, to
carry out onsite testing and to implement any required modifications
before certifying that thecontrolled plant is fully operational.

In this section we have focused on steps 2, 3, 4 and 5, and in this section we


will presentthree case studies which demonstrate many of the ideas and
practical techniqueswhich can be used in these steps. The case studies are
not meant to produce thebest controller for the application considered but
rather are used here to illustrate aparticular technique from the section.

In case study , a helicopter control law is designed for the rejection of


atmosphericturbulence. The gust disturbance is modeled as an extra input to
an S/KSHmixed-sensitivity design problem. Results from nonlinear
simulations indicatesignificant improvement over a standard S/KSdesign. For
more information on theapplicability of Hcontrol to advanced helicopter
flight, the reader is referred toWalker and Postlethwaite (1996) who describe
the design and ground-based pilotedsimulation testing of a high performance
helicopter flight control system.
This case study is used to illustrate how weights can be selected in H
mixed sensitivity design, and how this design problem can be modified to
improve disturbance rejection properties.

Problem description

In this case study, we consider the design of a controller to reduce the


effects ofatmospheric turbulence on helicopters. The reduction of the effects
of gusts is veryimportant in reducing a pilots workload, and enables
aggressive manoeuvers tobe carried out in poor weather conditions. Also, as
a consequence of decreasedbuffeting, the airframe and component lives are
lengthened and passenger comfort isincreased.

The design of rotorcraft flight control systems, for robust stability and
performance,has been studied over a number of years using a variety of
methods including:
Hoptimization (Yue and Postlethwaite, 1990; Postlethwaite and Walker,
1992);eigenstructure assignment (Manness and Murray-Smith, 1992;
Samblancatt et al.,1990); sliding mode control (Foster et al., 1993); and H2
design (Takahashi, 1993).The H controller designs have been particularly
successful (Walker et al., 1993),and have proved themselves in piloted
simulations. These designs have usedfrequency information about the
disturbances to limit the system sensitivity butin general there has been no
explicit consideration of the effects of atmospheric
turbulence. Therefore by incorporating practical knowledge about the
disturbancecharacteristics, and how they affect the real helicopter,
improvements to the overallperformance should be possible.We will
demonstrate this below.
The nonlinear helicopter model we will use for simulation purposes was
developed atthe Defence Research Agency (DRA), Bedford (Padfield, 1981)
and is known as theRationalized Helicopter Model (RHM). A turbulence
generator module has recentlybeen included in the RHM and this enables
controller designs to be tested on-linefor their disturbance rejection
properties. It should be noted that the model of thegusts affects the
helicopter equations in a complicated fashion and is self-containedin the
code of the RHM. For design purposes we will imagine that the gusts
affectthe model in a much simpler manner.
We will begin by repeating the design of Yue and Postlethwaite (1990) which
usedan S/KSH mixed sensitivity problem formulation without explicitly
consideringatmospheric turbulence.We will then, for the purposes of design,
represent gusts as aperturbation in the velocity states of the helicopter
model and include this disturbanceas an extra input to the S/KSdesign
problem. The resulting controller is seen tobe substantially better at rejecting
atmospheric turbulence than the earlier standardS/KSdesign.

The Helicopter Model

The aircraft model used in our work is representative of the Westland Lynx, a
twin-enginedmulti-purpose military helicopter, approximately 9000 lbs gross
weight,with a four-blade semi-rigid main rotor. The unaugmented aircraft is
unstable, andexhibits many of the cross-couplings characteristic of a single
main-rotor helicopter.In addition to the basic rigid body, engine and actuator
components, the model alsoincludes second order rotor flapping and coning
modes for off-line use. The modelhas the advantage that essentially the
same code can be used for a real-time pilotedsimulation as for a workstation-
based off-line handling qualities assessment.
The equations governing the motion of the helicopter are complex and
difficult to formulate with high levels of precision. For example, the rotor
dynamics areparticularly difficult to model. A robust design methodology is
therefore essential forhigh performance helicopter control. The starting point
for this study was to obtain

State Description
Pitch attitude
Roll attitude
p Roll rate (body-axis)
q Pitch rate (body-axis)
Yaw rate
vx Forward velocity
vy Lateral velocity
vz Vertical velocity
_

Table: Helicopter state vector

an eighth-order differential equation modelling the small-perturbation rigid


motionof the aircraft about hover. The corresponding state-space model is
where the matrices A;B and C for the appropriately
scaled system are available overthe Internet as described in the preface. The
8 state rigid body vector x is given in the Table. The outputs consist of four
controlled outputs

Together with two additional (body-axis) measurements

The controller (or pilot in manual control) generates four blade angle
demands which are effectively the helicopter inputs, since the actuators
(which are typically modelledas first order lags) are modeled as unity gains
in this study. The blade angles are

The action of each of these blade angles can be briefly described as follows.
The main rotor collective changes all the blades of the main rotor by an
equal amount and so

roughly speaking controls lift. The longitudinal and lateral cyclic inputs
change themain rotor blade angles differently thereby tilting the lift vector to
give longitudinaland lateral motion, respectively. The tail rotor is used to
balance the torque generatedby the main rotor, and so stops the helicopter
spinning around; it is also used to givelateral motion. This description, which
assumes the helicopter inputs and outputs aredecoupled, is useful to get a
feeling of how a helicopter works but the dynamics areactually highly
coupled. They are also unstable, and about some operating pointsexhibit
non-minimum phase characteristics.

We are interested in the design of full-authority controllers, which means that


thecontroller has total control over the blade angles of the main and tail
rotors, and isinterposed between the pilot and the actuation system. It is
normal in conventionalhelicopters for the controller to have only limited
authority leaving the pilot to closethe loop for much of the time (manual
control). With a full-authority controller, thepilot merely provides the
reference commands.

One degree-of-freedom controllers as shown in Figure are to be designed.


Notice that in the standard one degree-of-freedom configuration the pilot
reference commands r1 are augmented by a zero vector because of the rate
feedback signals.These zeros indicate that there are no a priori performance
specifications on y 2 =[ p q ]T.

Figure 1: Helicopter control structure (a) as implemented, (b) in the standard one
degree-of-freedom configuration

Hmixed-sensitivity design
We will consider theH mixed-sensitivity design problem illustrated in Figure
2.
Figure 2: S/KSmixed-sensitivity minimization

, an additional weight W3,W1 and W2 are selected as loopshapingweights


whereas W3 is signal-based. The optimization problem is to find astabilizing
controller K to minimize the cost function

This cost was also considered by Yue and Postlethwaite (1990) in the
contextof helicopter control. Their controller was successfully tested on a
piloted flightsimulator at DRA Bedford and so we propose to use the same
weights here. Thedesign weights W3,W1 and W2 were selected as

The reasoning behind these selections of Yue and Postlethwaite (1990)


issummarized below.

Selection of W1(s):For good tracking accuracy in each of the controlled


outputsthe sensitivity function is required to be small. This suggests forcing
integral actioninto the controller by selecting an s-1shape in the weights
associated with thecontrolled outputs. It was not thought necessary to have
exactly zero steady-stateerrors and therefore these weights were given a
finite gain of 500 at low frequencies.(Notice that a pure integrator cannot be
included in W1 anyway, since the standardH optimal control problem would
not then be well posed in the sense that thecorresponding generalized plant
P could not then be stabilized by the feedback
controllerK). In tuningW1 it was found that a finite attenuation at high
frequencieswas useful in reducing overshoot. Therefore, high-gain low-pass
filters were used inthe primary channels to give accurate tracking up to
about 6 rad/s. The presence ofunmodelled rotor dynamics around 10 rad/s
limits the bandwidth of W1. With fourinputs to the helicopter, we can only
expect to independently control four outputs.Because of the rate feedback
measurements the sensitivity function S is a six bysix matrix and therefore
two of its singular values (corresponding to p and q) arealways close to one
across all frequencies. All that can be done in these channels isto improve
the disturbance rejection properties around crossover, 4 to 7 rad/s, andthis
was achieved using second-order band-pass filters in the rate channels of W1.

Selection of W2(s):The same first-order high-pass filter is used in each


channelwith a corner frequency of 10 rad/s to limit input magnitudes at high
frequenciesand thereby limit the closed-loop bandwidth. The high frequency
gain of W2 can beincreased to limit fast actuator movement. The low
frequency gain of W2 was set toapproximately -100 dB to ensure that the cost
function is dominated by W1 at lowfrequencies.

Selection of W3 (s):W3 is a weighting on the reference input r. It is chosen


tobe a constant matrix with unity weighting on each of the output commands
anda weighting of 0.1 on the fictitious rate demands. The reduced weighting
on therates (which are not directly controlled) enables some disturbance
rejection on theseoutputs, without them significantly affecting the cost
function. The main aim of W3 is to force equally good tracking of each of the
primary signals.

For the controller designed using the above weights, the singular value plots
of Sand KSare shown in Figures 3(a) and 3(b). These have the general shapes
andbandwidths designed for and, as already mentioned, the controlled
system performedwell in piloted simulation. The effects of atmospheric
turbulence will be illustratedlater after designing a second controller in which
disturbance rejection is explicitlyincluded in the design problem.

Disturbance rejection design

In the design below we will assume that the atmospheric turbulence can be
modelledas gust velocity components that perturb the helicopters velocity
states vx, vy and vz by d = [ d1 d2 d3 ]T as in the following equations. The
disturbed system istherefore expressed as
Figure 3: Singular values of S and KS (Sand KSS=KS design)

whereG(s) = C(sI-A)-1B, and Gd(s) = C(sI- A)-1Bd. The design problemwe will solve is
illustrated in Figure4. The optimization problem is to find astabilizing
controller K that minimizes the cost function

r
which is the H norm of the transfer function from d to z. This is easily
cast intothe general control configuration and solved using standard
software. Notice that ifwe set W4 to zero the problem reverts to the
S/KSmixed-sensitivity design of theprevious subsection. To synthesize the
controller we used the same weights W1, W2and W3 as in S/KSdesign, and
selected W4 = I, with a scalar parameterused to emphasize disturbance
rejection. After a few iterations we finalized on = 30 . For this value of ,
the singular value plots of S andKS, see Figures 5(a)and 5(b), are quite
similar to those of the S/KSdesign, but as we will see in thenext subsection
there is a significant improvement in the rejection of gusts.
Also,sinceGdshares the same dynamics as G, and W4 is a constant matrix, the
degree ofthe disturbance rejection controller is the same as that for the S/KS
design.

Figure 4:
Disturbance rejection design

Figure 5: Singular values of S and KS (disturbance rejection design)


Conclusions
The two controllers designed were of the same degree and had similar
frequencydomain properties. But by incorporating knowledge about
turbulence activity intothe second design, substantial improvements in
disturbance rejection were achieved.The reduction of the turbulence effects
by a half in heave velocity, pitch attitude androll attitude indicates the
possibility of a significant reduction in a pilots workload,allowing more
aggressive manoeuvers to be carried out with greater precision.Passenger
comfort and safety would also be increased.The study was primarily meant
to illustrate the ease with which information aboutdisturbances can be
beneficially included in controller design. The case study alsodemonstrated
the selection of weights in H mixed-sensitivity design. To read howthe
Hmethods have been successfully used and tested in flight on a Bell 205
flyby-wire helipcopter, see Postlethwaite et al. (1999) and Smerlas et al.
(2001).

References

George Stephanopoulos, Chemical Process and Control An introduction to theory and


Practice Prentice Hall, Englewood Cliffs, NJ.

Doyle, J. C. and G. Stein (1981). Multivariable feedback design: concepts for a classical/
modern synthesis. IEEE Transactions on Automatic Control 26, 416.

Ogata, K. Modern Control Engineering Prentice Hall, Englewood Cliffs, NJ.

Morari, M. and E. Zafiriou Robust Process Control Prentice Hall International Ltd, London,
UK

Potrebbero piacerti anche