NASA TECHNICAL NOTE

F_-

N A S A TN D-3404
d -

I
W

t-'

t-'

s. gi
I

A METHOD OF TRAJECTORY OPTIMIZATION BY FAST-TIME REPETITIVE COMPUTATIONS

by Rodney C. Wingrove,James S. Ruby, and D. Francis Ames Keseurch Center Moffett Field, Cali$
'I

1'.

: .
.I

.,
.y

NATIONAL AERONAUTICS A N D SPACE ADMINISTRATION

WASHINGTON,

05;. .:;;

0;.

,A.P.R./~ i s a s
1'

- 9

$!

TECH LIBRARY KAFB, NM

IIlil11 1lulI 11111I l 11 0130190

NASA TN D-3404

A METHOD O F TRAJECTORY OPTIMIZATION BY FAST-TIME REPETITIVE COMPUTATIONS By Rodney C. Wingrove, J a m e s S. Raby, and D. F r a n c i s C r a n e Ame s Research Center Moffett F i e l d , Calif.

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

For sale

by the Clearinghouse for Federal Scientific and Technical Information Springfield, Virginia 22151 Price $0.30

A METHOD OF TRAJECTORY OPTIMIZATION BY FAST-TIME

RFPETITIVE COMmTTATIONS By Rodney C . Wingrove, James S. Raby, and D. F r a n c i s Crane Ames Research Center S U M M A R Y This r e p o r t p r e s e n t s a p e r t u r b a t i o n method of computing o p t i m t r a j e c t o r i e s wherein c o n t r o l impulse response f u n c t i o n s are determined by fast-time r e p e t i t i v e computations of t h e s t a t e e q u a t i o n s . This method does not r e q u i r e t h e s o l u t i o n of t h e a u x i l i a r y set of a d j o i n t e q u a t i o n s u s e d w i t h o t h e r p e r t u r b a t i o n methods. The mechanization of t h i s computing method on a h y b r i d computer i s d i s cussed and an a p p l i c a t i o n t o t h e s t e e p e s t descent o p t i m i z a t i o n of r e e n t r y t r a j e c t o r i e s i s p r e s e n t e d . I n t h i s example, t h e v e h i c l e i s t o a r r i v e a t a d e s i r e d range, t h e h e a t i n p u t t o t h e v e h i c l e i s t o b e minimized, and t h e con trol i s t o remain w i t h i n s p e c i f i e d c o n s t r a i n t s . Gptii-mm t r a j e c t o r i e s f o r t h i s example could b e o b t a i n e d i n about 8 minutes of computing time.
INTRODUCTION

It i s important f o r space v e h i c l e t r a j e c t o r i e s t o b e n e a r optimum i n t h e sense t h a t some q u a n t i t y i s e i t h e r maximized or minimized. For example, i n r e e n t r y t h e t r a i e c t o r y t o d e s i r e d t e r m i n a l c o n d i t i o n s i s n e a r optimum when t h e t o t a l aerodynamic h e a t i n g i s a minimum. This r e p o r t w i l l consider a method f o r f i n d i n g t h e Lime h i s t J r i e s of n o n l i n e a r c o n t r o l s t h a t correspond t o optimum t r a . i e c t o r i e s . S e v e r a l p e r t u r b a t i o n methods, such as t h e c a l c u l u s of v a r i a t b n s , a p p l i c a t i o n s of Ihe maximum p r i n c i p l e , and d i r e c t s t e e p e s t descent, have been considered f o r s o l v i n g t h i s c o n t r o l o p t i m i z a t i o n problem. Reference 1 c o n t a i n s a good review of t h e s e v a r i o u s methods and r e f e r e n c e 2 g i v e s s e v e r a l analog and d i g i t a l computing techniques f o r implementing them.

I n p r i n c i p l e , each of t h e s e techniques should g i v e s a t i s f a c t o r y r e s u l t s , b u t it has been found t h a t f o r many t r a j e c t o r y problems t h e computer mechani z a t i o n s are cumbersome and r e q u i r e programs t h a t a r e d i f f i c u l t f o r e n g i n e e r s t o formulate. The computing method t o be r e p o r t e d h e r e i n w a s i n v e s t i g a t e d i n an attempt t o a l l e v i a t e t h e s e d i f f i c u l t i e s and t o provide a more d i r e c t way of computing o p t i m i z a t i o n s o l u t i o n s .
I n previous o p t i m i z a t i o n s t u d i e s u s i n g p e r t u r b a t i o n techniques t h e computations have involved t h e dynamic s o l u t i o n of two s e t s of e q u a t i o n s : (1) n o n l i n e a r s t a t e e q u a t i o n s and (2) l i n e a r a d j o i n t e q u a t i o n s . The method t o b e r e p o r t e d h e r e i n d i f f e r s i n t h a t o n l y t h e s o l u t i o n of t h e n o n l i n e a r s t a t e

e q u a t i o n s i s used. The response of given f u n c t i o n s ( e . g . , t e r m i n a l e r r o r o r q u a n t i t y t o b e optimized) t o a c o n t r o l impulse i s determined along t h e t r a j e c t o r y by fast-time r e p e t i t i v e computations r a t h e r t h a n by a s o l u t i o n of t h e a d j o i n t e q u a t i o n s . Since a u x i l i a r y a d j o i n t e q u a t i o n s are n o t needed, t h e i n v e s t i g a t o r should u n d e r s t a n d t h e o p t i m i z a t i o n p r o c e s s more e a s i l y ; a l s o t h e computer program should be s i m p l e r . However, t h i s new a l t e r n a t e computing method does r e q u i r e many s o l u t i o n s o f t h e s t a t e e q u a t i o n s . This t a s k of com p u t i n g a l a r g e number of dynamic s o l u t i o n s i s i d e a l l y s u i t e d t o high-speed r e p e t i t i v e h y b r i d computation as w i l l b e considered h e r e i n . This r e p o r t w i l l p r e s e n t one a p p l i c a t i o n o f t h i s computing technique; t h a t of t r a j e c t s r y o p t i m i z a t i o n u s i n g t h e s t e e p e s t d e s c e n t method ( r e f s . 3 and 4 ) . The mechanization of t h i s method on a h y b r i d computer w i l l be d i s c u s s e d and r e s u l t s w i l l be p r e s e n t e d t o i l l u s t r a t e t h e u s e of t h i s procedure i n t h e o p t i m i z a t i o n of r e e n t r y t r a j e c t o r i e s . For t h e i n t e r e s t e d r e a d e r appendix A i l l u s t r a t e s t h e r e l a t i o n s h i p of t h e impulse response f u n c t i o n s computed i n t h i s r e p o r t t o t h e s o l u t i o n s o b t a i n e d w i t h t h e a d j o i n t e q u a t i o n s and t o t h e maximum p r i n c i p l e of o p t i m i z a t i o n . This appendix a l s o p r o v i d e s a background f o r under s t a n d i n g t h e s t e e p e s t descent o p t i m i z a t i o n e q u a t i o n s .

NOTATION The f o l l o w i n g n o t a t i o n i s u s e d i n t h e body of t h e t e x t .

Additional

symbols a r e d e s c r i b e d as t h e y are i n t r o d u c e d i n t h e appendixes.

L D
n c o n t r o l value of l i f t - d r a g r a t i o number of s t o r a g e p o i n t s i n c o n t r o l t i m e h i s t o r y

t tf
to

t ime
f i n a l t inie i n i t i a l time

At
U

t i m e increment of c o n t r o l impulse
1 function ccritr :

AU

c o n t m l impulse c o s t f u n c t i o n a t f i n a l time change i n c o s t f u n c t i o n a t f i n a l time due t o c o n t r a 1 impulses a t time t s t a t e v a l u e a t f i n a l time

&(t)

d e s i r e d s t a t e value a t f i n a l t i m e
A\I'(t)

change i n s t a t e value a t f i n a l t i m e due t o c o n t r o l impulses a t t i m e

T h e method of s t e e p e s t d e s c e n t ( r e f s . 3 and 4) i s an i t e r a t i v e procedure t h a t has been used for optimizing t r a j e c t o r i e s . The p r o c e s s commences w i t h any nonoptimal t r a j e c t o r y from which a s l i g h t l y improved one i s derived. The improved t r a j e c t o r y i s t h e n u s e d as a new nominal t r a j e c t o r y , and t h e procedure i s r e p e a t e d u n t i l t h e optimum or n e a r l y optimum t r a j e c t o r y i s found.

General O u t l i n e The i t e r a t i o n i s as f o l l o w s : (1) E s t i m a t e a reasonable program t h a t n e a r l y s a t i s f i e s t h e t e r m i n a l c o n d i t i o n s for s p e c i f i e d i n i t i a l c o n d i t i o n s ; ( 2 ) determine impulse response f u n c t i o n s t h a t d e s c r i b e t h e e f f e c t s of s m a l l changes i n t h e c o n t r o l on t h e t e r m i n a l s t a t e and on t h e c o s t ( t h e q u a n t i t y t o b e minimized). These impulse response f u n c t i o n s , combined w i t h s t e e p e s t descent computations, i n d i c a t e t h e b e s t p J s s i b l e way of making s m a l l changes t o t h e c o n t r o l t o decrease t h e c o s t and s t i l l a r r i v e a t t h e end p o i n t ; (3) add t h i s change i n c o n t r o l t o t h e previous nominal c s n t r o l program. The r e s u l t i s a new t r a j e c t o r y .with a decreased c o s t ; ( 4 ) r e p e a t t h i s p r o c e s s u n t i l t h e r e e x i s t s o n l y a v e r y s m a l l change i n c o s t f o r each new t r a j e c t o r y , i n d i c a t i n g t h a t t h e c o n t r o l i s v e r y near a l o c a l optimum. A l i m i t value of t h e c o n t r o l may b e reached b e f o r e t h e c o s t i s completely minimized. I n t h i s c a s e , t h e p r o c e s s i s continued u n t i l t h e c o n s t r a i n t ( c o n t r o l l i m i t ) i s reached, s i n c e no f u r t h e r o p t i m i z a t i o n i s p o s s i b l e . The p r o p e r t i e s of s t e e p e s t descent o p t i m i z a t i o n have been documented i n many p r e v i a u s s t u d i e s ( e . g . , r e f s . 5 - 7 ) . Although t h i s meth.;d has been regarded as t h e most p r a c t i c a l i n many a p p l i c a t i o n s , t h e r e i s no guarantee t h a t it y i e l d s t h e a b s o l u t e optimum. That i s , f o r some i n i t i a l choices of t h e nominal t r a j e c t o r y , t h e f i n a l optimized t r a j e c t o r y may r e p r e s e n t o n l y a l o c a l optimum p a t h . Also, i n some a p p l i c a t i o n s , where t h e c o s t f u n c t i o n may be r e l a t i v e l y i n s e n s i t i v e t o c o n t r o l v a r i a t i o n s , a l a r g e number of i t e r a t i o n s may be necessary t o approach t h e optimum s o l u t i o n . Computation of Impulse RespJnse Functions To i l l u s t r a t e t h e computation of t h e impulse response f u n c t i o n s l e t t h e q u a n t i t y t o be minimized be noted as cp, t h e c s s t e v a l u a t e d a t t h e f i n a l t i m e . L e t t h e s t a t e variable a t t h e f i n a l t i m e be noted as \I' and l e t t h e d e s i r e d end-point value f o r t h i s b e denoted \I'd. F i g u r e 1 i l l u s t r a t e s t h e manner i n which t h e i n f l u e n c e of s m a l l c o n t r o l changes on cp and \I' are c a l c u l a t e d i n t h i s r e p o r t . The e q u a t i o n s of motion are f i r s t s o l v e d w i t h a c o n t r o l change, a p o s i t i v e c o n t r o l impulse a t t i m e t ,

superimposed upon t h e nominal c o n t r o l . During t h e next s o l u t i o n of t h e motion e q u a t i o n s , a n e g a t i v e c o n t r o l impulse o f t h e same magnitude i s i n s e r t e d a t t i m e t . The change i n c o s t , 4, and t h e change i n t e r m i n a l state, A@, are d e r i v e d f r o m t h e s e t w o s o l u t i o n s . I n a s i m i l a r manner t h e impulse response f u n c t i o n s can be p r o g r e s s i v e l y determined at s u c c e s s i v e times along t h e t r a j e c t o r y , and t h e technique by which t h e y a r e determined i s t h e most important f e a t u r e o f t h i s computing method. The computation of t h e f u l l h i s t o r y o f @ ( t ) and A $ ( t ) f o r t h e same c o n t r o l impulse a t d i f f e r e n t t i m e s along t h e t r a j e c t o r y i s termed one " i t e r a t i o n " s i n c e it corresponds to t h e p r e v i o u s o p t i m i z a t i o n s t u d i e s where one i t e r a t i o n w i t h t h e a d j o i n t e q u a t i o n s w a s u s e d t o compute e s s e n t i a l l y t h e s e same f u n c t i o n s along t h e t r a j e c t o r y . Appendix A shows t h e r e l a t i o n s h i p o f t h i s experimental d e t e r m i n a t i o n t o t h e s t a n d a r d d e t e r m i n a t i o n u s i n g a d j o i n t equa t i o n s . These e x p e r i m e n t a l impulse response f u n c t i o n s a r e shown t o be d i r e c t l y r e l a t e d t o t h e well-known "Green's f u n c t i o n s . I'

S t e e p e s t Descent Optimization Equations The impulse response f u n c t i o n s a r e u s e d i n t h e s t e e p e s t d e s c e n t technique t o modify t h e c o n t r o l toward t h e optimum i n t h e f o l l o w i n g manner (see appen dix A).

The g a i n s Q and Kq a r e c o n s t a n t s f o r each i t e r a t i o n . The g a i n % weights t h e impulse response f u n c t i o n f o r t h e c o s t ; i t s s i g n i s n e g a t i v e i n o r d e r t o decrease t h e c o s t . The magnitude o f % i s determined e x p e r i m e n t a l l y f o r each problem. Too l a r g e a g a i n may cause i n s t a b i l i t y i n t h e convergence procedure, while t o o s m a l l a g a i n may extend t h e time o f convergence. The g a i n K q must be c a l c u l a t e d f o r each i t e r a t i o n s o t h a t t h e t e r m Kq Aq(t) w i l l account f o r t e r m i n a l displacement due t o t h e optimizing term, Kcp & p ( t ) , and any t e r m i n a l displacement e r r o r f r o m t h e p r e v i o u s i t e r a t i o n w i l l be c o r r e c t e d . The f o r m u l a t i o n f o r c a l c u l a t i n g K,J i s as f o l l o w s :
S m a l l changes, SJr, i n t h e t e r m i n a l s t a t e , $, due t o s m a l l changes, 6 u ( t ) , i n c o n t r o l can b e approx.imated by:

64f= 2 A : At

J
%

6 u (t ) A J r (t )d t
~

where A u i s t h e h e i g h t of each c o n t r o l impulse and A t i s t h e time i n t e r v a l of each c o n t r o l impulse. S u b s t i t u t i n g Q ( t ) + K,,, A $ ( t ) f r o m (1) f o r 6u, w e have :

6$ = 2 nu 1 A t Jt'
t0

(3)

Solving f o r

and I.etting -&$ = $d

$ (prevj.ous t e r m i n a l e r r o r ) we o b t a i n :

to J W S t e e p e s t descent optimization t e r m

<

to J v Terminal e r r o r correction t e r m The a c t u a l cal

This g i v e s t h e g e n e r a l form of t h e steepest descent e q u a t i o n s . c u l a t i o n s are next considered i n more d e t a i l .

HYBRID COMPUTER MECHANImTION

The mechanization of t h e o p t i m i z a t i o n procedure on a high-speed r e p e t i t i v e analog computer i s p r e s e n t e d i n f i g u r e 2. Figure 2 ( a ) i s t h e flow diagram and f i g u r e 2 ( b ) i l l u s t r a t e s t h e l o g i c u s e d i n a u t o m a t i c a l l y r e g u l a t i n g t h e problem. The mechanization c o n s i s t s of an analog computer program f o r s o l v i n g t h e t r a j e c t o r y e q u a t i o n s ; l o g i c r e q u i r e d t o c o o r d i n a t e t h e procedure; and a s e r i a l memory s t o r a g e u n i t f o r s t o r i n g t h e nominal c o n t r o l program. The s e r i a l memoryunit i s c o n t i n u o u s l y d r i v e n by counter p u l s e s (Logic no. 1 ) . The output of t h e s e r i a l memory i s t h e nominal c o n t r o l t i m e h i s t o r y w i t h n p o i n t s t h a t i s u s e d along w i t h t h e a p p r o p r i a t e c o n t r o l impulse, t o s o l v e t h e t r a j e c t o r y e q u a t i o n s . These equations are s t a r t e d a t t h e s p e c i f i e d i n i t i a l c o n d i t i o n s w i t h Logic no. 2 and stopped when t h e t r a j e c t o r y reaches t h e s p e c i f i e d end p o i n t w i t h Logic no. 3. The f i n a l v a l u e s of t h e c o s t quan t i t y , cp, and s t a t e q u a n t i t y , $', are s t o r e d a t t h e end of each r u n a s i n d i c a t e d by Logic n38. 4 and 5. The p z s i t i v e o r negative c o n t r o l impulse i s added t o t h e nominal c o n t r o l i n p u t w i t h Logic nos. 6 and 7 , r e s p e c t i v e l y . Logic no. 8 i n s e r t s t h e modifying c o n t r o l (Kcp @ + K ,, Aq) i n t o t h e s e r i a l memory. This p r s c e d u r e r u n s i n e s s e n t i a l l y a continuous manner; t h a t i s , one p o i n t o u t o f t h e n p o i n t s i n t h e nominal c o n t r o l h i s t o r y i s updated a f t e r each two r e p e t i t i v e computations, and a f t e r 2n r e p e t i t i v e computations (one i t e r a t i o n ) , every p o i n t i n s t o r a g e has been modified and t h e p r o c e s s i s r e p e a t e d . For each i t e r a t i o n t h e g a i n s Kcp and Kq are h e l d as c o n s t a n t s . A s p r e v i o u s l y men t i o n e d , t h e value of Kcp determines t h e r e l a t i v e speed and s t a b i l i t y of t h e convergence a n t 3 t h e optimum. The corresponding value of Kq t o b e u s e d w i t h each new i t e r a t i o n i s c a l c u l a t e d b y e q u a t i o n ( 4 ) as a f u n c t i o n of t h e t e r m i n a l e r r o r from each p r e v i o u s i t e r a t i o n ($d - $) and as a f u n c t i o n of t h e following two i n t e g r a t e d q u a n t i t i e s from each p r e v i o u s i t e r a t i o n :

and

The values f o r e q u a t i o n s ( 5 ) and (6) were computed as i n t e g r a l s a v e r t h e t i m e p e r i o d from t = to t o t = t f . The t i m e to w a s r e p r e s e n t e d b y a l o g i c s i g n a l a t t h e f i r s t r e p e t i t i v e computation i n an i t e r a t i o n c y c l e and t h e t i m e tf w a s r e p r e s e n t e d b y a l o g i c s i g n a l at t h e l a s t computation i n a m iteration c y c l e . It should b e n o t e d t h a t during t h o s e p u t s of t h e t r a j e c t o r y when t h e c o n t r o l w a s a t a c o n s t r a i n t l i m i t , no f u r t h e r o p t i m i z a t i o n w a s p o s s i b l e , and t h e i n t e g r a t i o n of e q u a t i o n s ( 5 ) and ( 6 ) w a s t h e r e f o r e n o t c a r r i e d o u t during t h o s e times. This type of computer mechanization w i l l b e i l l u s t r a t e d i n more d e t a i l f o r t h e f o l l o w i n g example problem.

APPLICATION TO FEEITTRY TRAJECTORY OPTIMIZATION

Statement of t h e Problem The problem t o b e i l l u s t r a t e d i n t h i s s e c t i o n i s t h a t of determining t h e time h i s t o r y of t h e v a r i a t i o n of l i f t - d r a g r a t i o ( c o n t r o l L/D) t h a t must b e flown f o r a v e h i c l e r e t u r n i n g i n t o t h e e a r t h ' s atmosphere s o t h a t : The t o t a l h e a t i n g load t o t h e v e h i c l e i s minimized. The v e h i c l e arrives a t a d e s i r e d d e s t i n a t i o n . The c o n t r o l remains w i t h i n s p e c i f i e d c o n s t r a i n t s .

Mechanization The equations of motion, p r e s e n t e d i n appendix.B, were f o r a p o i n t mass i n p l a n a r motion 3ver a s p h e r i c a l n o n r o t a t i n g e a r t h . The v e h i c l e c h a r a c t e r i s t i c s and f l i g h t c o n d i t i o n s were t h o s e f o r a manned capsule r e t u r n i n g from e a r t h orbit. I n i t i a l c o n d i t i o n s were: Altitude
Velocity
F l i g h t - p a t h angle
Range t o d e s t i n a t i o n
F i n a l stopping c o n d i t i o n s were: Altitude

76.3 lan 7.63 km/s 1 . 8 ' 1609 Ism

(250,000 f t ) (25,000 f p s )

(1000 m i )

30.48

~nn

(ioo,ooo f t )

C o n t r o l l i m i t s were:

o <L<

0 . 5

The main hardware elements u s e d i n t h e h y b r i d computer mechanization were: Hardware elements Analog computer P a r a l l e l d i g i t a l logic units Track and s t o r e a m p l i f i e r s D i g i t a l d e l a y l i n e memories1 ( w i t h D/A and A/D c o n v e r t e r s ) Program t a s k S o l u t i o n of t r a j e c t o r y e q u a t i o n s Logic c o n t r o l of program S t o r a g e o f end-point v a l u e s S t o r a g e of c o n t r o l time h i s t o r y

The 64-word d i g i t a l s e r i a l memory u n i t (1.3 b i t s p e r word) w a s accessed w i t h t h e f a s t e s t allowable c o u n t e r rate (0.002 s e c ) . A complete 64-word c y c l e w a s t h e n a v a i l a b l e e v e r y 0.128 second. To a l l o w a complete s o l u t i o n o f t h e t r a j e c t o r y e q u a t i o n s w i t h i n 0.128 second, t h e analog computer w a s time s c a l e d a t 3750 t o 1. Results
A s e r i e s of computer runs f o r t h i s problem i s i l l u s t r a t e d i n f i g u r e 3. F i g u r e 3 ( a ) p r e s e n t s t h e d e t a i l s of each r e p e t i t i v e t r a j e c t o r y computation and f i g u r e 3 ( b ) p r e s e n t s t h e d e t a i l s Jf t h e o v e r a l l convergence onto t h e optimum nominal c o n t r o l . F i g u r e 3 ( a ) shows j u s t a p o r t i o n o f i t e r a t i o n no. 0 as p r e sented i n f i g u r e 3 ( b ) .

I n t h e upper t r a c e of f i g u r e 3 ( a ) t h e c o n t r o l impulses a r e superimposed upon t h e i n i t i a l nominal c o n t r o l . Each c o n t r o l impulse had a magnitude o f L/D = 20.25 and a time increment of one c l o c k p u l s e (0.002 s e e ) . This c o n t r o l impulse w a s chosen because it gave v a r i a t i o n i n t h e f i n a l range and h e a t l o a d on t h e o r d e r o f f 5 p e r c e n t . The range and i n t e g r a t e d h e a t l o a d along each of t h e r e p e t i t i v e t r a j e c t o r i e s a r e p r e s e n t e d along w i t h t h e f i n a l v a l u e s as t h e y a r e s t o r e d w i t h t r a c k and s t o r e a m p l i f i e r s . The d i f f e r e n c e between t h e s e s t o r e d q u a n t i t i e s f o r each two p a i r s o f subsequent runs i s A@ r e p r e s e n t i n g t h e range impulse response f u n c t i o n s and &p representing the heat load impulse response f u n c t i o n s . I n f i g u r e 3 ( b ) t h e f i r s t 10 i t e r a t i o n s ( e a c h i t e r a t i o n c o n s i s t s of 128 r e p e t i t i v e computations) of t h e converging o p t i m i z a t i o n procedure are i l l u s t r a t e d along w i t h t h e f i n a l i t e r a t i o n . During t h e convergence procedure t h e range i s seen t o vary s l i g h t l y about t h e d e s i r e d value o f 1609 km (1000 m i l e s ) . The h e a t l o a d i s shown t o b e reduced about 10 p e r c e n t during t h e f i r s t t e n i t e r a t i o n s and diminished t o about 12 p e r c e n t from t h e o r i g i n a l w i t h t h e f i n a l (optimum) c o n t r o l v a r i a t i o n .
A s e r i e s of t r a c k and s t o r e a m p l i f i e r s could a l s o have been u s e d for this stxage.

The modifying c o n t r o l shown i n t h e f i g u r e i s t h e sum Q nCp + Kq A $ . For t h i s s e r i e s of runs a c o n s t a n t v a l u e of E;lp = - 2 . 5 ~ 1 0 - ~ [ u n i tof s (L/D)/(J/m2)] w a s found t o , a l l o w a f a i r l y r a p i d convergence while maintaining program s t a b i l i t y . The v a l u e of K$ w a s c a l c u l a t e d by e q u a t i o n ' ( 4 ) t o b e t h a t value f o r each i t e r a t i o n such a s ' t o a l l o w convergence i n t h e s t e e p e s t descent manner. I n t h e lower t r a c e of f i g u r e 3 ( b ) t h e nominal c o n t r o l i s recorded as it i s r e a d o u t of s e r i a l memory every 128 +1 counter p u l s e s ( w i t h Logic no. 8 ) . This g i v e s a convenient time h i s t o r y t o show t h e manner i n which t h e c o n t r o l has been modified during each i t e r a t i o n . The c o n t r o l i s seen t o be l i m i t e d w i t h i n 0 < L/D < 0.5. This w a s achieved simply by l i m i t i n g t h e output of t h e s e r i a l memory t o w i t h i n t h e s e v a l u e s .

As can be seen, t h e optimum c o n t r o l v a r i a t i o n f o r t h i s case i s a bangbang c o n t r o l . With t h e s t e e p e s t descent method, it i s u s u a l l y found t h a t n e a r optimum c o n t r o l can be achieved i n t h e f i r s t few i t e r a t i o n s , b u t t h a t t o "square up t h e corners'' and achieve f u l l optimum c o n t r o l a number of f u r t h e r i t e r a t i o n s (on t h e o r d e r of 20 t o 50) are r e q u i r e d .
Convergence and S t a b i l i t y Considerations One of t h e important a s p e c t s of any o p t i m i z a t i o n scheme i s t h e a b i l i t y t o converge, w i t h i n a reasonable time, onto t h e optimum s o l u t i o n . For t h e p a r t i c u l a r method i n th'is r e p o r t it has been p o i n t e d o u t t h a t t h i s convergence p r i m a r i l y depends upon choosing t h e p r o p e r v a l u e of t h e g a i n $. I n t h e example l e s s than 2 . 5 ~ 1 0 - ~ problem, it w a s found t h a t using any value of [ u n i t s of (L/D) /( J / m 2 ) ] r e s u l t e d i n smooth convergence; however, t h e convergence time (which w a s p r o p o r t i o n a l t o 1 / % ) became long. For i n i t i a l values of g r e a t e r t h a n twice t h e aforementioned value t h e convergence became u n s t a b l e , t h a t i s , t h e modifying 6 c o n t r o l became s o l a r g e as t o change d r a s t i c a l l y t h e s t a t e v a r i a b l e from t h e i r nominal f i n a l v a l u e s .

KCP

It w a s found t h a t , as t h e optimum c o n t r o l w a s approached ( a f t e r about 10 i t e r a t i o n s ) , t h e value of % could be i n c r e a s e d and convergence s t a b i l i t y maintained, because i n t h e s e examples t h e c o n t r o l approached bang-bang and o n l y s m a l l changes were p o s s i b l e n e a r t h e s a t u r a t i o n limits. The value of K$ i n t h e s e c a s e s could be i n c r e a s e d t o about 10 times t h e aforementioned value, b u t i n c r e a s i n g it much f a r t h e r (without analog v o l t a g e s c a l i n g changes) would a l l o w extraneous computer n o i s e t o b e magnified t o a p o i n t where it caused n o t a b l e random f l u c t u a t i o n s i n t h e computations.

For a reasonable value of g a i n , such as t h a t used f o r t h e example problem, t h e time t o converge t o a n e a r optimum s o l u t i o n ( 1 1 i t e r a t i o n s ) w a s about 3 minutes, and t o a f u l l optimum s o l u t i o n (30 i t e r a t i o n s ) , about 8 minutes. F u r t h e r changes i n t h e s e convergence times, of course, depend upon s e v e r a l f a c t o r s . For i n s t a n c e , t h e convergence time i n t h i s computing setup w a s i n p r o p o r t i o n t o n2, where n i s t h e number of p o i n t s d e s c r i b i n g t h e c o n t r o l time h i s t o r y (64 p o i n t s f o r t h e case c i t e d ) . Also t h e allowable s o l u t i o n r a t e s of t h e computer elements d i r e c t l y a f f e c t t h e convergence time. The continuing development and use of high-speed computing elements w i l l c e r t a i n l y r e s u l t i n convergence times s m a l l e r t h a n t h e time c i t e d .

The r e s u l t s o b t a i n e d b y t h i s computing method appear s a t i s f a c t o r y f o r e n g i n e e r i n g purposes; however, t h e u s u a l disadvantages o f analog computation are i n h e r e n t w i t h t h i s method. These disadvantages are p r i m a r i l y concerned w i t h t h e extraneous n o i s e i n t h e computations and t h e a b s o l u t e accuracy ( o n l y t o w i t h i n about 1 p e r c e n t ) of analog computer.
CONCLUDING

REMARKS

This r e p o r t has p r e s e n t e d a p e r t u r b a t i o n method of computing optimum t r a j e c t o r i e s . The technique u s e s fast-time r e p e t i t i v e computations i n d e t e r mining c o n t r o l impulse response f u n c t i o n s and r e q u i r e s o n l y t h e dynamic s o l u t i o n o f t h e s t a t e e q u a t i o n s ; whereas o t h e r p e r t u r b a t i o n computing t e c h n i q u e s have r e q u i r e d t h e s o l u t i o n of a d d i t i o n a l a d j o i n t e q u a t i o n s . A h y b r i d computer w a s used i n applying t h e method t o t h e s t e e p e s t descent o p t i m i z a t i o n of r e e n t r y t r a j e c t o r i e s . Mechanizing t h e computer f o r t h i s type of problem w a s r e l a t i v e l y simple, and near Q p t i m m t r a j e c t o r i e s could be o b t a i n e d i n about 3 minutes of computing t i m e and full optimum t r a j e c t o r i e s i n about 8 minutes. The advantage o f t h e technique o u t l i n e d here over a l t e r n a t i v e techniques i s t h a t t h e i n v e s t i g a t o r need n o t be familiar w i t h or u s e an a u x i l i a r y s e t of l i n e a r a d j o i n t e q u a t i o n s f o r t h e o p t i m i z a t i o n . This technique &ses, however, r e q u i r e a l a r g e number of dynamic s o l u t i o n s of t h e s t a t e e q u a t i o n s , b u t t h i s computing t a s k appears p r a c t i c a l w i t h t h e high-speed r e p e t i t i v e computation procedure p r e s e n t e d i n t h i s r e p o r t . Ames Research Center N a t i o n a l Aeronautics and Space Administration Moffett F i e l d , C a l i f . , J a n . 24, 1966

APPENDIX A

RELATIONSHIP O F

EXPERIMENTAL IMPULSE RESPONSE FUNCTIONS

TO ADJOIN" SOLUTIONS AND THE MAXIMUM PRINCIPLJT

A d j o i n t S o l u t i o n of Impulse Response F u n c t i o m

L e t the s t a t e e q u a t i o n s b e n o t e d as

where x ( t ) i s a v e c t o r of s t a t e v a r i a b l e s , u ( t ) i s t h e c o n t r o l v a r i a b l e and i s a v e c t o r of known f u n c t i o n s of x ( t ) and u ( t )

The a u x i l i a r y a d j o i n t e q u a t i o n s can b e noted as

where A i s a v e c t o r of i n f l u e n c e f u n c t i o n s and (af/ax)T i s t h e t r a n s p o s e of t h e matrix d e s c r i b i n g l i n e a r motions about t h e nominal p a t h of x ( t ) and

u(t)
It i s w e l l known t h a t any s m a l l change i n t h e c o n t r o l q u a n t i t y along t h e nominal p a t h w i l l determine a change 6(p i n any q u a n t i t y f i n a l t i m e as follows:
6u(t) cp a t t h e

+ r e p r e s e n t s a s o l u t i o n of t h e a d j o i n t e q u a t i o n s w i t h t h e boundary where & c o n d i t i o n s a t t h e f i n a l t i m e of

The q u a n t i t y

A ( a f / d u ) w i t h i n t h e i n t e g r a l i s kn3wn a s Green's f u n c t i 3 n .

1 0

Experimental Determination of Z q u l s e Response Functions

The method used i n t h e t e s t .of t h i s r e p o r t f o r f i n d i n g a change p e r t u r b the c o n t r o l e x p e r i m e n t a l l y i n t h e f o l l o w i n g manner:

6 c q

is to

ti

L J !
t0

,Nominal control
. . . .=

tf

With two s e q u e n t i a l dynamic s o l u t i o n s of the s t a t e e q u a t i o n s u s i n g f i r s t a p o s i t i v e c o n t r o l i m p u l s e and t h e n a n e g a t i v e c o n t r o l impulse, t h e f o l l o w i n g i s avai l a b l e :

From e q u a t i o n (A3) we can write t h e change 26q = Ap, f o r a s m a l l c o n t r o l , 6u = nu, a c t i n g over a s m a l l t i m e i n t e r v a l A t , as f o l l o w s :

2ep =
=

( . p T af z ) 2 nu

At

This then r e p r e s e n t s t h e correspondence between t h e impulse response f u n c t i o n s c a l c u l a t e d i n t h e text and t h o s e s o l v e d by t h e a d j o i n t s o l u t i o n . Greens f u n c t i o n e v a l u a t e d a t any t i m e , t , along t h e t r a j e c t o r y can b e noted as

and

R e l a t i o n s h i p t o t h e Maximum P r i n c i p l e

The maximum p r i n c i p l e ( r e f . 8) s t a t e s that a n e c e s s a r y c o n d i t i o n f o r a minimum (maximum) of t h e c o s t f u n c t i o n i s t h a t t h e Hamiltonian be maximized (minimized) w i t h r e s p e c t t o t h e c o n t r o l at a l l t i m e s . The Hamiltonian can be w r i t t e n as

H = A f

(A91
11

where t h e t r a n s v e r s a l i t y c o n d i t i o n must b e s a t i s f i e d a t t h e f i n a l time,

and i s a Lagrange m u l t i p l i e r c o n s t a n t chDsen s3 t h a t t h e t e r m i n a l con s t r a i n t i s s a t i s f i e d . The boundary c o n d i t i o n s on t h e a d j o i n t e q u a t i o n s a t t h e f i n a l time are :

Now t o determine i f H i s minimized w i t h r e s p e c t t o t h e c o n t r o l we can t a k e t h e p a r t i a l d e r i v a t i v e of H w i t h r e s p e c t t o u:

Or, n o t i n g t h e correspondence between e q u a t i o n s (Ah) and (All), we can write

T where &,(af/au) i s Green's f u n c t i o n f o r t h e c o s t and functionTfor t h e terminal constraint.

A,(af/au)

i s Green's

R e c a l l i n g t h e correspondence between t h e a d j o i n t s o l u t i o n for Green's f u n c t i o n and t h a t determined e x p e r i m e n t a l l y , we have t h e f o l l o w i n g :

T h i s , then, r e p r e s e n t s t h e r e l a t i o n s h i p between t h e e x p e r i m e n t a l l y det emnine d impulse r e sponse f u n c t i o n s and t h e Hamiltonian. It i s i n t e r e s t i n g t h a t t h e maximum p r i n c i p l e can be a p p l i e d through t h i s r e l a t i o n s h i p without any need f o r s o l v i n g t h e a d j o i n t e q u a t i o n s . S t e e p e s t Descent Equatlons

>
a

The g r e a t e s t change, @, i n obtained (ref.

cp

f o r a given value of
t0

6u2(t)dt

is

3) when

12

where Kcp and % are c o n s t a n t s . T h i s i s t h e s t e e p e s t descent ( o r a s c e n t ) direction t o the mini" (or maximum) cp. When t h e r e are no s t a t e o r c o n t r o l c o n s t r a i n t s , t h e s t e e p e s t descent procedure converges toward t h e n e c e s s a r y c o n d i t i o n s f o r an optimum s o l u t i o n as p r e v i o u s l y noted

Q(t> 4- -q
where i n t h e s t e e p e s t descent e q u a t i o n s , with the terminal constraint s a t i s f i e d .

a+>=
K q / %

( A16 1
on t h e o p t i m s o l u t i o n

= 7,

APPENDIX B
REENTRY T W C T O R Y EQUATIONS
The f o l l o w i n g e q u a t i o n s were programmed on t h e analog computer for t h e example c o n s i d e r e d i n this r e p o r t . These s i m p l i f i e d e q u a t i o n s were d e r i v e d f o r f l i g h t w i t h i n t h e atmosphere and t h e primary assumptions i n c l u d e : a s p h e r i c a l n o n r o t a t i n g e a r t h , s m a l l f l i g h t - p a t h a n g l e s , and a c o n s t a n t g r a v i t y term. The d e r i v a t i o n of t h e s e s i m p l i f ied e q u a t i o n s and t h e i r a p p l i c a b i l i t y have been c o n s i d e r e d i n a number of r e p o r t s . See f o r i n s t a n c e r e f e r e n c e 9. The e q u a t i o n s are

$ =

st'
t0

V dt

where

L D
h

c o n t r o l value of l i f t - d r a g r a t i o altitude, m h o r i z o n t a l v e l o c i t y , m/s f i n a l range, m

cp
P

t o t a l h e a t i n p u t , J/m2
atmosphere d e n s i t y , 1.225 e radius from earth center, -h/7160

%/m3

r
g

6.43~10~ m
9.8 m/s2

l o c a l gravitational acceleration, drag l o a d i n g , 0.004 mz/kg

(?)
14

REFERENCES

1 . Leitmann, George, ed.:

Optimization Techniques.

Academic Press, 1962.

2. Balakrishnan, A. V. ; and Neustadt, Lucien W., eds. : Computing Methods in Optimization Problems. Academic Press, 1964.
I.

3. Bryson, Arthur E. ; and Denham, Walter F. : A Steepest-Ascent Method for Solving Optimum Programming Problems. Raytheon Rep. BR 1303, 1961. Also J. Appl. Mech., vol. 29, no. 2, June 1962, pp. 247-257.

4 . Kelley, H. J. : Gradient Theory of Optimal Flight Paths. ARS J., 30, no. 10, Oct. 1960, pp. 947-954.
5. Bryson, A. E.; Denham, W. F.; Carroll, F. J.; and
Mikami,

vol.

tion of Lift or Drag Programs That Minimize Reentry Heating. space Sei., vol. 29, no. 4, April 1962, pp. 420-430.

K.: Determina J. Aero

6. Blanton, H. Elmore, ed. : Three-DimensionalTrajectory Optimization Study. Pt. 1 - Optimum Programming Formulation. NASA CR-57030, 1964.
(Supersedes Aero. Sys. Div. Rep. ASD-TDR-62-29?and Raytheon Rep. Br-1759-1).

7. Hague, D. S. : Three-Degree-of -Freedom Problem Optimization Formulation. FDL-TDR-64-1,pt. 1, V O ~ . 3, Oct. 1964.

8. Pontryagin, L. S., et al.: The Mathematical Theory of Optimal Processes.
Wiley and Sons, 1962.

9. Chapman, Dean R.: An Approximate Analytical Method for Studying Entry

Into Planetary Atmospheres. NASA TR R-11, 1959.

C , o n t r o I im pu Ise
Control, U

71

State va Iue

cos

t
t0

1 1 1 1 1 1

tf to Trajectory t i me 1 1 1 1 1 1 1 1 Computer t i m e

tf

1 1 1 1 1 1

F i g u r e 1,- I l l u s t r a t i v e time h i s t o r y of r e p e t i t i v e analog computations.

I.

c.

91 (Counter)

Analog

f 6
impulse O

I
L

I I

-Control impulse

3+2 (Start)

I I I

+
I----I
#3
(Stop)
I

I I

----Dotted

lines represenl logic signals shown in figure 2 ( b )

(a) Flow diagram.

Figure 2,- Mechanization of the optimization procedure.

#2

-"

L
(Stop)

#3

Solution time

J
I

#A

(itoie

(Store)

#5

#7

(- Control )

I
I

impulse)

#8
(Modify)

I
( b ) Problem l o g i c .

Figure 2.

Concluded.

rc

.5 Control L .25 -

n
'

/Control

impulse Nominal control

il
~ y , I y

0-

Final range,$, IO3 km

2 IW-

Final heat Ioad,4, 20

-i
0.1 Second of computer time

( a ) D e t a i l s of r e p e t i t i v e t r a j e c t o r y computations.
Figure 3.- Recorded time h i s t o r i e s for r e e n t r y t r a j e c t o r y optimization; minimum h e a t with t e r m i n a l range constraint

? Iu
Iu
P

Final range, IO3 km

+
IO

Final heat load, $ 107 Joule

m2

0-

I O Seconds of computer time

2-

Modifying -.I 6 control O .I -

%- 1 1

( b ) Details of o v e r a l l convergence procedure.

Figure

3.- Concluded,

The aeronautical and space activities of the United States shall be conducted so as to contribute . . . to the expansion of human knowl edge of phenomena in the atmosphere and space. The Administration shall provide f o r the widest practicable and appropriate dissemination of information concerning its actiuities and the results thereof.
-NATIONAL AERONAUTICS AND SPACE ACT OF 1958
\ i

NASA SCIENTIFIC A N D TECHNICAL PUBLICATIONS

TECHNICAL REPORTS: Scientific and technical information considered important, complete, and a lasting contribution to existing knowledge. TECHNICAL NOTES: Information less broad in scope but nevertheless existing knowledge. of importance as a contribution TECHNICAL MEMORANDUMS: Information receiving limited distri bution because of preliminary data, security classification, or other reasons. CONTRACTOR REPORTS: Technical information generated in con nection with a NASA contract or grant and released under NASA auspices. TECHNICAL TRANSLATIONS: Information published in a foreign language considered to merit NASA distribution in English. TECHNICAL REPRINTS: Information derived from NASA activities and initially published in the form of journal articles. SPECIAL PUBLICATIONS: Information derived from or of value to NASA activities but not necessarily reporting the results .of individual NASA-programmed scientific efforts. Publications include conference proceedings, monographs, data compilations, handbooks, sourcebooks, and special bibliographies.

Details on the availability o f these publications may be obtained from:

SCIENTIFIC AND TECHNICAL INFORMATION DIVISION

N AT1 0 NA L AERO N AUT1 CS A N D SPACE A D M I N ISTRATI 0 N

Washington,

D.C. PO546