Vous êtes sur la page 1sur 7

Interactive design optimization of dynamic systems

C. H. TSENG A N D J. S. A R O R A
Optimal Design Laboratory, College o f Engineering, The University o f Iowa, Iowa City,
IA 52242-1593, U.S.A.

This paper describes an interactive environment


for dynamic response optimization problems. It is
shown that with proper interactive facilities and
designer intervention, the optimal design process
can be more flexible, efficient and effective in
solving practical design problems. The process can
be initiated with a rough design model for the
problem. As the iterations progress and insight is
gained into the problem, the model can be interactively rehned, without stopping the process, to
obtain the hnal solution. Two example problems a system dynamics problem and a control problem
- are used to study and show advantages of the
designer interaction. It is concluded that interactive optimum design capabilities offer more opportunities to the designer to obtain better designs.

The present paper demonstrates the use of an interactive design optimization environment for the dynamic
response application, including control of systems. It is
shown that not only the efficiency of the optimal design
process can be improved but also that some difficult
minimum time optimal control problems can be solved
with an alternate formulation. During the interactive
session, the designer can refine the model for the
problem, based on the insights gained, to obtain a
more accurate solution. Therefore interaction in the
computer-aided design optimization process can be very
useful for practical applications.
In Section 2, a general mathematical model for
optimal design and control of dynamic systems is
defined. A prototype interactive environment for the
selected application is described in Section 3. Example
problems are presented in Section 4 and conclusions are
given in Section 5.

2. DEFINITION OF THE PROBLEM


1. INTRODUCTION
The purpose o f this paper is to describe and demonstrate advantages of designer interaction for dynamic
response optimization o f systems. Interactive design
optimization is a computer-aided environment where
the designer has an opportunity to interact with the
iterative process and make design decisions. Such an
environment needs a sophisticated and flexible software
system. Therefore, a prototype system for dynamic
response applications is developed and briefly described
in the paper.
Optimal design of practical systems is a fairly
complex process requiring integration of several software components and numerical procedures. This
integration is described for the dynamic response
applications. Sufficient details are given, so that using
the sample application, other applications can be developed.
A certain amount of uncertainty usually exists in
computer implementation of numerical algorithms.
Therefore, it is prudent to interactively monitor the
computational process and change the course of calculations if necessary. A strong case for interactive design
optimization and their benefits for practical design
applications are also discussed in detail in Refs 1-3.

Accepted April 1990. Discussion closes April 1991.

1990 Computational Mechanics Publications

A general formulation that encompasses dynamic


response as well as control problems is defined as follows: find the design variable vector b E R g, the control
variable vector u ( t ) ~ R m, and terminal time tl that
minimize a cost functional or the performance index
(PI):
~bo= go(b, z(tf), tf) +

I"

Fo(b, u(t), z(t), t) dt

(1)

to

subject to the state (or dynamic) equation

to <~t <~tf

~(t) = f(b, u(t), z(t), t),

(2)

with the initial conditions. Z(to)--h(b), the functional


constraints

~ = g~(b,z(tf),tf)+

F~(b,u(t),z(t),t) dt
to

= 0; c = 1
r'
~<0; a = r ' + 1 ..... r

(3)

and time dependent point-wise constraints

~z(b,u(t),z(t),t)

= 0;/3 = 1 , . . . , q '
~ < 0 ; / 3 = q ' + l ..... q

(4)

In the preceding definition, z(t)~ R n is the state variable that is assumed to be a continuous function of
time, and u(t)ER m is assumed to be bounded and
measurable on the interval [to, tf], where to and tf are
the initial and terminal times; to is kept fixed and tf is

Adv. Eng. Software, 1990, Vol. 12, No. 4

167

treated as a design variable. The differential equations


for the system in equation (2) are written in general first
order form, where a ' . ' denotes the time derivative of
the state variable. It should be noted that even though
z(t) appears linearly, these equations can be nonlinear.
Equation (4) represents time dependent constraints on
the state and control variables as well as explicit bounds
on the design variables. Functions go, F0, g~, f, F~, and
~ are assumed to be at least twice differentiable.
The model defined in equations (1) to (4) is quite general. The cost functional can be time, control effort,
error from a nominal response, or any other functional
of state, control or design variables. Inequality or
equality constraints can be imposed on the state or
control variables at any time or over any time interval.
The problem is difficult to solve due to its transient
dynamic nature as well as point-wise state variable constraints in equation (4).
Numerical methods to discretize and solve the foregoing problem were recently evaluated in Refs 4 and 5.
These included treatment of time dependent constraints, integration of differential equations, interpolation of variables, and methods of design sensitivity
analysis and integral evaluations. Several test problems
from dynamic response optimization and optimal
control literature were used there to evaluate the
performance of the numerical procedures for design
optimization. It was shown that the computational
effort varied substantially depending on the numerical
procedures used. It is shown in this paper that efficiency
of the dynamic response optimization process can be
improved by starting the process with inaccurate but
efficient numerical procedures. As improved designs are
obtained, the numerical procedures can be interactively
changed to obtain a more accurate solution.

Data Entry I

Design

UserSupplied
Problem
Definition

I/11
Interactive
Graphics

Output:
Final Design

IDESIGN
Fig. 1. Conceptual layout
optimization environment

of

interactive

design

IDESIGN

3. I N T E R A C T I V E E N V I R O N M E N T FOR
DYNAMIC RESPONSE OPTIMIZATION

To create an interactive design optimization environment for the dynamic systems modeled by equations (1)
to (4), proper hardware and software are needed. In
addition, several software components must be properly integrated. Here we describe a possible environment for the dynamic response applications. Using this
prototype, an interactive design optimization environment for other applications can be created.
The Apollo workstations, the general purpose design
optimization shell IDESIGN 1'2'6, and other software
components are used to create the present environment.
Figure 1 shows the overall arrangement of the system.
An application is installed into IDESIGN through usersupplied subroutines that define problem functions and
their gradients. Two levels of interaction are possible,
as shown in the conceptual layout of the system in
Fig. 1. The first level is the general purpose interaction
shown on the left side. This type of interaction is not
tied to any application and uses only general optimization terminology. It can be easily incorporated in the
general purpose optimizer. The domain-specific interaction is shown on the right side of the figure. This type
of interaction uses the domain-specific terminology and
commands, and must be developed for each application. This has been done for the present application, as
will be described later in the paper. The software con-

168

Adv. Eng. Software, 1990, Vol. 12, No. 4

7-

;-

Fig. 2. Conceptual arrangement of user-supplied subroutines for dynamic response optimization application
taining dynamic response optimization capabilities is
called OCP (Optimal Control Problem).
The interactive capabilities of IDESIGN and their
usage are described in detail in Refs 1 and 2, so they will
not be repeated here. Additional capabilities for the
dynamic response application will be discussed. The
capability for optimum design of dynamic systems
modeled by equations (1) to (4) is installed into
IDESIGN by creating the following four subroutines:
USERMF for the cost function, USERCF for the
constraint functions, USERMG for the cost function
gradient, and USERCG for the constraint gradients.
Through these routines, application-dependent software, such as differential equation solvers, interpolation routines, gradient evaluation routines and integral
evaluation routines, are called to evaluate problem
functions and their gradients. Domain-specific interactive commands can be added in these subroutines.
Figure 2 shows the arrangement of the user-supplied

TBFIT:

Subroutine to calculate coefficients of the


interpolating function.
TGVAL:
Subroutine to evaluate the interpolated
function and its gradient.
SIMPSN:
Subroutine to evaluate an integral by
Simpson's method.
GAUSS:
Subroutine to evaluate an integral by the
Gauss-Legendre quadrature formula. It is
assumed that the number of Gaussian
points, N, is some power of 2, and that
2 ~< N ~< 256.
DDERKF: Subroutine to integrate first order differential equations by Runge-Kutta-Fehlberg
method.
DDEABM: Subroutine to integrate first order differential equations by Adams-BashforthMoulton predictor-corrector method.
DDEBDF: Subroutine to integrate first order
differential equations by backward differentiation formula of order one through
five.

Fig. 3. Structure o f C T R L M F module f o r cost function evaluation

subroutines for this application. Each subroutine has a


model that controls the use of various numerical procedures. Figure 3 shows some details of the CTRLMF
module for the cost function evaluation. This module is
described in the sequel to give some details of the OCP
software. Other control modules have similar structure.
Various subroutines of the CTRLMF module, shown
in Fig. 3, have the following functions (a rectangular
block indicates a control routine; lines having double
arrows that connect various blocks indicate bidirectional data flow):
CTRLMF:

CTRLPR:

CTRLFT:

ANALYS:

CTRLF:

INGSOL:
PREPRO:
ZOFN:
DIFSOL:

Control subroutine that calculates the performance index (cost function), partitions
the incore array, and keeps track of its
usage.
Subroutine to read additional input data
for the dynamic and control optimization
problem.
Subroutine to interpolate the control function with respect to the design variable
vector.
Subroutine to set up initial conditions for
the state equation and to calculate state
trajectories; it calls several other routines
to accomplish its function.
Control subroutine to calculate coefficients
of the interpolation function, function
values, or their first derivatives at a given
grid point. Subroutine calls TBFIT and
TGVAL.
Subroutine to control method for numerical integration.
Subroutine to calculate value of the
problem functions, defined in equations
(1), (3) and (4), at a specified time t.
Initial state z(t0) evaluation subroutine.
Subroutine to control the solution method
for differential equations.

Methods for treatment of time dependent constraints


and their sensitivity analysis are implemented in the
gradient evaluation modules through subroutines
CTRLMG and CTRLCG. Additional input data for a
specific dynamic response optimization problem are
entered through the subroutine CTRLMF in the first
iteration. These data and the variable names used for
them are described in the following paragraphs (some
of the variables described below are used later in the
presentation of numerical results). Note that some of
the data items can be also changed during the interactive design session.
1. The optimal control problem parameters:
IRP: Number of equality functional constraints
IR: Number of inequality functional constraints
IQ: Number of time dependent constraints
2. The next data line contains information about
initial and terminal times:
TO: Initial time
TF: Terminal time
3. The next data block contains nine parameters for
specifying numerical procedures (Ref. 4 may be
consulted for more details of the numerical procedures):
IDINTR: Index for the interpolation scheme for
the state variables
0 Zero order piece-wise polynomial interpolation function
1 First order piece-wise polynomial interpolation function
3 Cubic spline with natural boundary condition
at the end points
IDINTU: Index for the interpolation scheme for
the control variables (same as IDINTR)
IDDIFF: Index for the differential equation
solver
1 Runge-Kutta-Fehlberg method (DDERKF)
2 Adams-Bashforth-Moulton formula
(DDEABM)
3 Backward differentiation formula
(DDEBDF)

Adv. Eng. Software, 1990, Vol. 12, No. 4

169

IDINTG: Index of the integration scheme


1 Simpson's rule
2 Gaussian quadrature formula
INTGPT: Number of integration points (note
that I N T G P T must be a power of 2, and 2 ~<
I N T G P T ~< 256 for Gaussian quadrature formula)
IDPTWC: Index for the treatment of time dependent constraints 4
1 Impose constraints at all grid points (conventional approach)
2 Impose constraints at the local max-points
among the grid points
3 Impose constraints at the local maximum
points (worst-case approach)
4 Hybrid treatment combining conventional
and worst-case approaches
IDMETH: Index for the method of design
sensitivity analysis 4
0 Hybrid method combining direct differentiation and adjoint variable methods (HYB)
1 Direct differentiation method (DDM)
2 Adjoint variable method (AVM)
IDTIMS: Index for terminal time
0 ti is fixed
1 ti is free (free time problem)
IDFORC: Index for input force
0 Control forces are unknown
1 Discretized control force given in a tabular
form
2 Control force known in an explicit functional
form and the user needs to supply subroutine
FORCE
4. The next data line contains information about the
number of equations and grid points
NEQ:
Number of state equations
MU:
Number of control forces
NVCTRL: Total number of control variables in
the design variable vector
NGPT:
Number of grid points for state variable in [to, ts]
N G P T U : Number of grid points for control
function in [to, ts]
Additional interactive capabilities are implemented in
the C T R L M F routine to control the number of time
grid points, the method for treatment of dynamic constraints 4, the interpolation scheme, the acceptable error
in generating the state trajectory, and some other input
data. These capabilities will be demonstrated in the next
section.

variables (NGPTU), smaller number of integration


points (INTGPT), larger error tolerance for the differential equation solver (RELERR - relative error and
ABSERR - absolute error), and conventional treatment of dynamic constraint (IDPTWC = 2) or smaller
number of worst case constraints (IDPTWC = 3 where
only a few more severely violated worst cases are
included in finding the search direction). In other
words, more accurate solutions are possible at the
expense of efficiency. Therefore, the idea of designer
interaction for dynamic response problems is to select
more efficient numerical schemes at the beginning of the
optimization process. Once this process has converged,
it can be switched to more accurate numerical procedures to obtain a better solution. This way, one can
obtain accurate optimum solutions more efficiently. In
addition, if the design process is not proceeding satisfactorily, it may be terminated to study and analyze the
problem formulation.

4.1 A five degree of freedom vehicle suspension


system 7
Figure 4 shows a five degree of freedom planar model
of a vehicle. The suspension system for the vehicle is to
be designed to minimize the extreme acceleration of the
driver's seat for a variety of vehicle speeds and road
conditions. Spring constants kl, k2 and k3 and damping
coefficients ci, c2 and c3 of the system are chosen as
design parameters. The design of the system calls for
constraints on the relative displacements between the
chassis and the driver's seat, and the chassis and the
f r o n t and the rear axle. Furthermore, it is desired to
constrain driver seat acceleration due to an additional
set of extreme road conditions. Lower and upper
bounds on the design variables are also placed.
A combination of two vehicle speeds and two road
profiles, shown in Fig. 5, is considered as input to the
vehicle model of the present example (v in Fig. 5 corresponds to the vertical displacement at the front tire).
This corresponds to Case 3 of the problem given in
Ref. 7. Detailed formulation for the problem and
design data are given there. The problem has ten first
order state equations and six dynamic inequality constraints for each road condition, so it is a moderate size
problem. There are six design variables (spring and
damper parameters). It is desired to minimize the

zl~__

,2t _

4. EXAMPLES AND RESULTS


Two examples that have been used extensively in the
recent literature - a five degree of freedom vehicle suspension system given in Ref. 7 and a minimum time
control problem given in Ref. 8 - are used to demonstrate the use of the interactive capability. Both
problems can be described by the mathematical model
given in equations (1) and (4).
Based on the numerical experience in Refs 4 and 5,
the following parameter values give good performance
with respect to efficiency but not accuracy: smaller
number of grid points for the state (NGPT) and control
170

Adv. Eng. Software, 1990, Vol. 12, No. 4

C.G,@ )+

~J z 5
m2g

c,

z'L
Cs~b

k2~>ljC2

.
f2())

jC4

,,I,)l_r_L"1

Fig. 4. Five degree of freedom vehicle model

v(y)

L .....

36o"
5"

:"

(a)

Pro, f l i t ,

N,~. I

v(y)

t
~.y

(b)

Profile

~n. 2

Fig. 5. Road surface profiles


Table 1. Interactive execution of vehicle suspension design problem
Stal~e her. Max. Vio. Con. Pm'a. Cost

lmeraetive Change
I D F r W C = 2; N G P T = 101:
INTGPT = 32
RELERR = 3.E-5; ABSERR = I.E-5
ACS = 0.5; A C V = 0.01

5.07E-01

....

1.986E+02
Design change
No. Value
5 2.500E+01
6 2.500E+01

1.23E-01

---

1.986E+02

I
8

1.18E-02 1.53E+00 1.986E+02


3.58E-03 3.18E-01 1.112E+02

by linear exttapolatlon
Suggest
Choice
5.449E+01 5.449E+01
8.000E+01 8.000E+01

Design changeby linear extrapolation


No. Value Suggest
Choice
1 1.000E+02 5.000E+01 5.000E+01
4 1.000E+01 5.000E+01 5.000E+01
5 5.449E+01 6.743E+01 6.743E+01
Optimum Solution is found
Re,sl~t f ~ m It~ttinn No. 8
1DITrWC = 4; N G P T = 1001;
I N ' l ~ l r r = 128
RELERR = 3.E-7; ABSERR = I.E-7;
ACV = 0.001; ACS = 0.01

S 3.58E-03 3.78E-01 1.112E+02


22 0.00
6.33E-03 9.914E+01

Optimum Solution is found

Numbea'ofuallsforfuuetiodaevalualion = 29
Total n u ~
of gradients evaluated
= 1568
CPU = 38528 seconds for APOLLO DN460
With worst case ~,~,at~em o f ' d y n a m i c ~ s t r a i n t s and batch environment [51
Cost function value = 9.894E+01
Numbeaofc~lls fc function evaluation = 60
Total number of gradients evaluated
= 159
CPU = 59464 seconds for APOLLO DN460
Q

maximum acceleration of the driver's seat over the specified time and the road conditions. The problem formulation can be transcribed into the form of equations
(1) to (4).
The problem is solved using a two phase process.
During the initial phase, design changes are made
according to the suggestions given by IDESIGN
without doing any optimization. The numerical
schemes are selected according to their performance
with respect to efficiency. During the second phase, the
numerical schemes are selected to obtain a more
accurate solution. The step-by-step procedure is illustrated in Table 1 and explained as follows:

Stage I: Inaccurate solution is obtained with the following parameter values: N G P T = 101, I N T G P T = 32,
RELERR=3.E-5
and A B S E R R = I . E - 5
(error
tolerance for integrating differential equations),
I D P T W C = 2 (the max treatment of dynamic constraints among the grid pointsa), ACV = 10070 (tolerance for constraint violation), and A C S = 0 . 5
(convergence parameter for IDESIGN).
Iteration 1 :
Maximum violation of constraints is large (50.770).
IDESIGN gives suggestions, without optimization,
for changing design variable numbers 5 and 6 to
improve constraint violations. After the suggested
design changes, the maximum violation is still large
(12.3070). Another suggestion is given for changing
the design variables 1, 4, and 5. The maximum violation is reduced to 1.2070 after the suggested changes
are made. At this stage, a command to obtain the
optimum solution is given.
Iteration 8:
Optimum solution is found with relaxed convergence
criteria. The minimum acceleration is about 12070
higher than the true optimum.
Stage 11." To obtain a more accurate solution, the
iterative process is restarted from iteration number 8
with the parameters and schemes as N G P T = 1001,
I N T G P T = 128,
R E L E R R = 3" E - 7,
ABSERR =
1.E-7,
IDPTWC=4
(the hybrid treatment of
dynamic constraint described in Refs 4 and 5), and
ACV = 0.1070 and ACS = 0.01 for IDESIGN.
Iteration 8:
The maximum violation is the same as in the first
stage. The command to obtain optimum solution is
given.
Iteration 22:
The optimum solution found is the same as the one
given in the literature 4'5'7.
Data for comparison between interactive and batch
solution procedures are given in Table 1. It can be seen
that the problem is solved more efficiently with the
interactive procedure. As seen in Table 1, the total
number of iterations and the CPU time are reduced
substantially compared with results given in Refs 4 and
5.
From the results for this example, we see that with
reasonable interactive changes, we can solve dynamic
response optimization problems more efficiently as well
as accurately.

4.2. Minimum time control problem 8


The minimum time control problems are difficult to
treat numerically. One difficulty is that since the total
time is a variable, it changes from one iteration to the
next. Therefore, if the number of grid points is kept
fixed, the grid size changes every iteration. This is particularly true in the initial stages of the optimization
process. Since the grid size is a variable, the constraint
functions may not be evaluated at the same point from
iteration to iteration. Due to this uncertainty in the
constraints and their gradients, a search direction for
design improvement may not exist, especially when the
Adv. Eng. Software, 1990, Vol. 12, No. 4

171

problem involves bang-bang control. The design


process can be slow, or fail to converge.
It turns out that the minimum time control problem
can be formulated and solved using an equivalent
minimum control effort problem. The problem is
solved for a fixed terminal time with the same state
equations, initial and terminal conditions, and the constraints, as for the minimum time control problem. The
minimum control effort problem is well behaved and
converges in a few iterations, especially when the time
interval is large enough. Once the optimum solution is
found, the time interval is interactively reduced and the
problem solved again. The process is continued until
the time interval is too small, making the problem
infeasible. This can be observed interactively. The terminal time and other parameters can be adjusted to
solve the problem again. The process is continued until
no further reduction in the time interval is possible.
The minimum time control problem, called the
bounded control of a double-integrator plant, is taken
from Ref. 8. The problem is to find a control function
u(t) for 0 <~ t <~ tf, to minimize the performance index

7,0

Stage

Max. Vio.

Conv. Para.

P1

1
2
3
4
9

5.0000E+O0
1.2410E-05
2.9292E-06
2.1550E-06
1.8534E-05

1.0000E+00
1.1959E-02
1.1557E-02
9.6765E-03
7.0102E-04

0.0000E+O0
|.6382E-01
1.6363E-01
1.6277E-01
1.6001E-OI

Optimal

I1

9
19

1.2500E+00
1.1030E-05

9.9785E-01
3.3716E-04

1.6001E-01
7.6930E-0 l

2.5
Optimal

II1

19
20
2I
22
23

4.0000E-01
3,7878E-01
3.7862E-01
3.7860E-01
3.6666E-01

9.1396E+00
1.2292E+03
5.5097E+01
5.2948E+00
7.8681E+00

7.6930E-01
8.3701E-01
8.6526E-01
8.6533E-01
8.9445E-01

2.0

Restart/19

IV

19
23

9.5994E-02
2.9773E-02

9.7985E-01
3.1322E+00

7.6930E-0!
9.0926E-01

2.4
Restart/19

19
23

4.9000E-02
2.6534E-07

3.5677E-01
3.1743E-04

7.6930E-01
8.4935E-01

2.45
Optim',d

VI

23
24

2.4745E-02
1.3464E-04

3.961 IE-01
5.3055E-04

8.4935E-01
9.1781E-01

2.425
Opdmal

VII

24
28

4.9858E-03
3.5331E-03

1.0743E-01
2.1537E-01

9.1781E-01
9.2441E-01

2.42
Restarl/24

VIII

24
28

1.9944E-03
8.8149E-04

4.2557E-02
1.0820E-01

9.1781E-01
9.2263E-01

2.423
RestarT24

IX

24
30

9.9563E-04
6.4601E-05

2.1182E-02
6.3473E-05

9A781E-01
9.2198E-01

2.424
Optimal

ts

subject to the state equations


zl = z2,

z2 = u(t)

with the initial conditions Z l ( 0 ) = 0 . 5 and zz(O)= 1.0,


the bounded terminal conditions
--0.5 <~ Z l ( t f ) <~ 0.5,

lu(t)l~<l.0

for

her.

tf
5.0

Number of calls for function evaluation = 87


Total number of gradients evaluated
= 3849
CPU = 10492 seconds for APOLLO DN460

Zz(tf)=O

and the bounded control as


O<~t<.tf

This problem is in the form of the general formulation


given in equations (1) to (4).
The interactive capabilities for solving minimum time
problems via minimum control effort formulation are
demonstrated with this example. Initially, a value for
the terminal time is selected and the optimization
process is started. The process is monitored interactively. If the maximum constraint violation remains
large and the value of the performance index increases,
or the convergence parameter has a large value in most
of the iterations, the process needs to be restarted with
a larger value for the time interval because the problem
is infeasible. Otherwise, the design process can converge very rapidly. The following data are kept fixed for
the entire interactive session (they can also be changed,
however): N G P T = 1 0 1 , N G P T U = 2 1 , R E L E R R =
3.E-7,
ABSERR=I.E-7,
I D T G P T = 6 4 ; cubic
spline piecewise interpolation scheme for the state and
control variables, Adams-Bashforth-Moulton differential equation solver, the hybrid method for design sensitivity analysis, and the hybrid treatment o f dynamic
constraints as described in Refs 4 and 5 are used. The
interactive procedure is illustrated in Table 2 and
explained as follows:
Stage I: Terminal time (tf) is selected as 5 seconds and
the control force u(t) is set to zero.
Iteration 1 :
Maximum constraint violation is 50070 and performance index is 0. The command to find optimum is
given.

172

Table 2. Interactive execution .[or minimum time control problem

Adv. Eng. Software, 1990, Vol. 12, No. 4

Iteration 9:
Optimum is found with the performance index (control effort) as 0.16.
Stage 11: Terminal time is reduced to 2.5 seconds and
the process restarted from Iteration 9.
Iteration 9:
Maximum violation is 125o70 and performance index
is 0.16.
Iteration 19:
Optimum is found with the performance index as
0.769.
Stage IlL" The terminal time is reduced to 2 seconds
and the process restarted from iteration 19.
Iteration 19:
Maximum violation is 4070, and the convergence parameter has large value, 9.14.
Iteration 23:
After 5 more trials, the constraint correction is very
slow and convergence parameter is still large. It is
concluded that the minimum control effort problem
with the terminal time as 2 seconds is infeasible.
Stage IV: Since the problem with the terminal time as
2 seconds in Stage III is not feasible, the terminal time
needs to be increased. It is increased to 2.4 seconds and
the process is restarted from iteration 19. After 5 more
trials, the behavior is almost the same as in Stage III.
Therefore, the problem is still unfeasible. The optimum
terminal time for the original minimum time control
problem is between 2.4 and 2.5 seconds. If the designer
is satisfied with this solution, the interactive process can
be terminated here.

Stages V - I X : The same procedure as in previous stages

is used to determine the minimum terminal time in


Stages V to IX. The final solution for the terminal time
is between 2.423 and 2.424 seconds. This result is the
same as given in Refs 5 and 8.

5. CONCLUDING REMARKS
Interactive use of a prototype system for o p t i m u m
design and control of dynamic systems is demonstrated.
Advantages of designer interaction are demonstrated
by solving two simple examples. During problem
execution, the system allows the designer to control the
optimization process, and the numerical procedures
for analysis and design sensitivity analysis. The five
degrees of freedom vehicle suspension problem is
solved more efficiently and accurately by the use of
interactive facilities. The number of iterations and the
C P U time are reduced considerably compared to those
with the automated environment. The minimum time
control problem is solved using the interactive facilities
and the minimum control effort formulation.
The present study and other recent results 1'2 show
that designer interaction during the optimization
process is quite beneficial. For example, the designer
can change the problem formulation without quitting
the iterative process. For the present examples, the
number of time grid points, the method o f treatment of
time dependent constraints, and the accuracy controls
for generating the state trajectories and the final design,
are changed interactively. This interactive environment
can be easily expanded to include more facilities as well
as other applications. For the finite element analysis
applications, one m a y start with a coarser discretization
and interactively refine the model at a later stage. This
way, practical solutions with desired accuracy will be
obtained more efficiently. In addition to the above, the
designer can monitor the process for a few iterations
and if errors occur or are suspected in the formulation
or implementation, the p r o g r a m can be terminated at
an early stage, saving time and money.
For large and complex design problems, the interactive process can be slow because each iteration requires
time-consuming analysis and design sensitivity analysis

o f the system. Thus interactive design optimization may


become a time-consuming job for the designer. One solution to this problem is to use a supercomputer at the
backend to handle the number-crunching problem after
interactive decisions have been made at the workstation.
In conclusion, designer interaction during the
optimization process is highly desirable and recommended for practical applications. However, further
research is needed to create a more sophisticated
interactive environment and fully exploit the facilities.
In this regard, knowledge-based expert system ideas
and artificial intelligence strategies can be quite
useful 9,1.

REFERENCES
1 Arora, J. S. and Tseng, C. H. Interactive design optimization,
Engineering Optimization, 1988, 13, 173-188
2 Arora, J. S. Introduction to Optimum Design, McGraw-Hill,
New York, 1989
3 Park, G. J. and Arora, J. S. Role of database management in
design optimization systems, J. of Aircraft, 1987, 24, (11)
745-750
4 Tseng, C. H. and Arora, J. S. Optimum design of systems for
dynamics and controls using sequential quadratic programming,
1989 AIAA J., 27 (12), 1793-1800
5 Tseng,C. H. and Arora, J. S. Optimal Design for Dynamics and
Control by using a Sequential Quadrature Programming Algorithm, Technical Report No. ODL-87.10, Optimal Design Lab-

7
8
9

oratory, College of Engineering, The University of Iowa,


U.S.A., IA 52242, 1987
Arora, J. S. and Tseng, C. H. IDESIGN User's Manual: Version
3.5, Technical Report No. ODL-87.1, Optimal Design Laboratory, College of Engineering, The University of Iowa, USA, 1A
52242, 1987
Haug, E. J. and Arora, J. S. Applied Optimal Design: Mechanical and Structural Systems, John Wiley and Sons, N.Y., 1979
Lee, A. Y. Dynamic optimization problems with bounded
terminal conditions, Journal of Optimization Theory and
Applications, 52, (i) 151-162
Baenziger, G. and Arora, J. S. Development of an artificial
intelligent nonlinear optimization expert system, Proceedings of
the 27th American Institute of Aeronautics and Astronautics,
Structures, Structural Dynamics and Materials Conference, San

Antonio, Texas, May 19-21, 1986, 66-77


10 Arora, J. S. and Baenziger, G., Uses of artificial intelligence in
design optimization, Computer Methods in Applied Mechanics
and Engineering, 1986, 54, 303-323

A d v . Eng. Software, 1990, Vol. 12, N o . 4

173

Vous aimerez peut-être aussi