Professional Documents
Culture Documents
Technion - Computer Science Department - Technical Report CS0055 - 1975
Technion - Computer Science Department - Technical Report CS0055 - 1975
by
z. Ritz and I.Amit
Technical Report No.55
November 1975
This report has been submitted for publication elsewhere and has been issued as a
I, I NTRODUCT! ON
Whether or not a hit will be scored upon firing a single A,A. round
trajectories of the target and the shell, among them the initial and instan
hand the values of the various parameters at the time of fire, there is no
assurance that aiming is indeed accurate, therefore it is customary to
move at constant speed along a straight I ine during the time of fire.
instrument.
2. THE SIMULATION-PROGRAM
the same for most conventional A.A. systems) must correctly simulate
both the gun-aiming process and the errors which are inherent to it.
toward the Theoretical Impact Point (TIP) defining the TIP to be the
point in space at which a properly directed projectile fired at the right
moment will strike the target, provided the estimates of target and shell
We divide the main errors appearing during this process roughly into
trajectories of target and shell and therefore lead to the wrong TIP-the
process due to mechanical I imitations of the gun system and ball istic
2.1 The target is spherical with ra*i~s r and moves at a constant speed
angles respectively of the target in relation to the gun (see Fig. 1).
We derive the above restriction from the evidence that aircraft which attack
-~-
point ground targets with conventional armamant, such as bombs and rockets,
attempt to fly at a constant speed and in a straight 1ine during the attack
stage, which is also the stage at which the A.A. batteries usually fire
accurately calculate the ball istic trajectory of a shell (see Macshane [5J).
For our purposes, equations (1) and (2) are sufficiently accurate, especially
2.3 The gun system estimates of target speed, dive and course anglesarasimulated
" "
as normally distributed random variables N(uo ,0 u2 ) , N(a 0 , 0a2 ) , N(epo'O~)
respectively, with the actual values of these parameters uo ' " "
ao' epo
as means and au, 0a' 0ep' the standard deviations, as input variables.
~ The elevation and prediction angles at which the shell is actually fired
2.7 A hit is scored when and if at any moment the centers of target and
shell are at a distance of less than r from each other.
A A
-+
Let y(t;v,Ct,S) be the instantaneous position vector of the shell at
time of
Technion shell);
- Computer Sciencelet w(t)- Technical
Department be the average
Report CS0055 -speed
1975 of the projectile along
A A
-+ .
- 7
-+ ""
Using (1), the formula for y(t;V,el,S) in the (X,Y,Z) system is:
-+ " "
y(t;V,el,S) = (w(t)tcoselsinS,w(t)tcOSelCOSS,w(t)tsinel) (2)
-+ " "
x(t;u,e,~) = (utcosesin~,d-utcosecos~,utsine) (3)
determine the target's trajectory are sampled form the given normal
Step 2: With the above estimates, utilizing formulas (1), (2) and (3),
the theoretical hit time to' and "ell and "a , are calculated
l
by solving the equation
-+ " "
Iy(t;vo,el,a) - -+ ""
x(t;u,e,~) I2 = 0, (4)
or exp 1 ic i t 1y :
where
Let n denote the total number of trials for a specific case, m the
total number of hits scored, and Pn the SP estimate of the hit-probabil ity
after n trials; then
Pn = mIn (])
where R.I is the minimal distance between target and shell in the i-th
trial.
00030
0.0251. o =0.033 Ct =0
e a Ct
o =0,01
Ct 0cj>=Oe
\
0.0201
~n
O.Oli
r<
j
0,0101
1I
:1
0.005;
L m 400
-+
800
--lI~-_._._._.
1200
,
1600
~
2000 2400 d
a
O· 009 j a=0~
C1<p=Oe
0.008[
0.007
0.006 j
0,,005J \
0.004 ~~=0,,02 .~
fe=O.03
~-~
0.003 f~="''''~
p=O.Ol
. . . .
"~
......."
""
0.002 : . e " "'-.. .
0001~~~
radian 0 0.3491 0.6981 1.0472 1.3963"
<Po in Radian
...,.
We let ~ denote these six parameters, that is to say,
...,. A l;, A A
~ = (v,a.,S,u,a,<j» .
...,.
2 -+ --+
N(C. ,0 .. ),
10 i
i = 1, ••• ,6; let z(t;Z;) and be the instantaneous
and she 11
-+ -+ A ~
z ( t ; l;) = -+
y(t;v,a., ) - -+ " "
x(t;u,a,<j» (10)
.3
-+
Q.(t;l;) =
-+ -+
II Z ( t; l;)ll = [ i =I 1z. (t; t) ~
I - *
( 11)
We assume
-+ -+
l:.!. T is a positive number such that for any value of l;, ( t ; t;) has
only a single local minimum in the time i nterva 1 [O,TJ. That is to say,
-+
if R(l;) denotes that minimum and t the time it occurs, then
l
R( +)
z:; = ~
(-+
t i ; 1;) < ~,
,
t ; +)
1; (12)
O~tl<T Vt,tE[O,TJ,t~tl
R(!o ) = O. ( 13)
-+
is a continuous function of the random vector (,;, therefore
Theorem:
( 14)
where
S2 2 + [w (t )/(l+kt )J2+
= u0 2u [w (t )/(l+kt )]cos(a +~) (16)
00 0 000 0 0 0
Eli s the e r ro r .
We assume, as did Cunningham [I], Frazer [2], and Lind [4J, that
from the dispersion of the intersection between shell trajectory and the TP.
(n ,n ) , with the center of target as the mean and (1,(1 as the respective
1 2 1 2
standard deviations.
p = J I N(O,0~)N(O,(1~)dnldn2' ( 18)
n2+n 2s r 2
1 2
If (11 = O
2
= (1 the hit-probabil ity is given by
( 19)
Alternatively, if
(21)
and
We find that taking aa - a<l> ' aa. "" aa and <I> - 0 is a sufficient
condition to cause a 1 = a2 . This condition defines a category of cases
where formula (22) can be used for hit-probabil ity estimation.
5. CONCLUSIONS
Some interesting relationships utilizing formulas (22) and (a14) are represented
the costs of changes in and are defined, we can calculate the quality of
the reductions in 0a and 0e which will most efficiently shift the AA system
from one indifference curve to a better one.
ACKNOWLEDGEMENT
The authers wish to express their thanks to Dr. A.Shafi for his valuable
remarks and suggestions.
O. 13 ~O. OOO~ ,
: : p~~\\
o. 1
0.05
O.oR
0.07
0.06
0005
0.04
\
0.03
0.0
0.01 \\
radian l - _......_ . - . _......-'-+l-_ _- - l l - - - . ; I - - - - -
radian 0.01 0.02 0.03 0.04 0.05 0.06 0.07
°a.
APPENDIX A
where
ll/3!) l la3)
i ,j , k= 1
6
E.
o
a) = (1/4!) . . l _ [olt R2(t l )lor;i or;}r;k 0I;R,] _.n. (I;n - r;no) . la4)
I ,J , k, R,- 1 n- I ,J ,k, R,
(a5)
standard deviation a,
x
then
(a6)
E{T} = L (a])
i=l
and
6
= (9 ·6/4 1) L [a 4 R2 (t )lar;~ar;~ ] a~a~ (a8)
i ,j=l 1 I J I J
-+ -+ -+ -+ -+
G(t; r;) = z(t; r;)·(dz(t; r;)/dt) = 0 (alO)
consequent 1Y ,
= 1"",6 , (all)
i = 1, ... ,6 . (a 12)
(a 13)
where
and
Q.E.D.
Since
and
A A A
and since, in effect, the standard-deviations of the angles a, S, 8
A
and $ are quite small, there are cases where E2 can be neglected,
2
and 1 can be estimated by
m
12 = E{T} . (aI4)
m