Beruflich Dokumente
Kultur Dokumente
=
= +
=
=
(1)
The corresponding variables are deIined in Table 1
|5|
.
Table1. Notation
Variable Description Units Variable Description Units
pitch angle rad Y liIt N
pitch
angular rate
rad/s
v
c
liIt
coeIIicient
1/rad
pitch angular
acceleration
rad/s
2
q
dynamic
pressure
N/s
2
trajectory
angle
rad S
reIerence
area
m
2
trajectory
angular rate
rad/s o
Iin
deIlection
rad
o
angle
oI attack
rad ay
normal
acceleration
m/s
2
m mass kg P thrust N
Variable a a a b b c J
Units s
-2
s
-2
s
-1
s
-1
s
-1
m m/s
Value +250 280 1.5 1.6 0.23 0.681 914.4
Wherein, the positive c value means that the accelerometer
is ahead oI the CG.
Thereby three possibilities exist:
a
Fig 1. Autopilot working procedure
A very general approach to the missile control
philosophy is to measure the angular rate
and normal
acceleration a
y
oI the missile. Wherein, the real measured
quantity oI an accelerometer includes the angular
acceleration
g
k
dc
k
a
k
Fig 2. Two-loop topology
u
vc
a
v
a c +
1
i
p
i
Ts
k
T s
+
PI compensator
Fig 3. Two-loop topology with PI compensator
u vc
a
v
a c +
i
s
dc
k
a
k
g
k
Fig 4. Three-loop topology
Rewritten in state space notation, the basic missile
plant is
m m m m
x x u v x u = + = + (2)
To be more speciIic, the short period dynamics are
shown as Iollows
to a two-loop autopilot
[ ]
T
x =
, [ ]
T
v
v a c = +
1
m
b
a a
=
(
(
m
b
a
(
=
(
0 1
m
Jb ca ca
(
=
(
0
m
ca Jb
=
(
(
to a two-loop autopilot with PI compensator
[ ]
T
x =
, [ ]
T
v
v a c = +
0 1
0 0 1 0
0
m m
b b
a a a
= =
( (
( (
( (
( (
0
0 0 1 0
1 1 0 0
m m
Jb ca ca ca Jb
= =
( (
( (
( (
( (
to a three-loop autopilot
[ ]
T
x =
, [ ]
T
v
v a c = +
0 1
0 0 1 0
0
m m
b b
a a a
= =
( (
( (
( (
( (
0
0 0 1 0
0 1 0 0
m m
Jb ca ca ca Jb
= =
( (
( (
( (
( (
The nominal static stable plant is a relatively Iast
lightly damped system. While let a
0, the plant is
relatively Iast unstable system, which yields the system
eigen equation can be decomposed into two aperiodic loops
with nearby absolute-values and opposite signs. Figure 5
shows the pure missile airIrame response.
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
a
y
m
/
s
2
t(s)
a
= 250m/s
2
a
= -250m/s
2
a
= 0
Fig 5. Pure airIrame response
To reveal the beneIit oI applying RSS technology on
missile, the relation between normal overload (n
y
) and
static stability (a
) can be set up as
3660 2008 Chinese Control and Decision Conference (CCDC 2008)
( ) sin
v
v
c qS P Y P
n
mg mg
+ +
= (3)
With moment relation on balance state, the equation is
given as
( )
v
v
c qS P a
n
mg a
+
= (4)
As other variables keep almost constant, the missile
normal overload will increase 3-4 times iI plant static
stability (a
= + + + (5)
where it has made the polynomial a real pole times a
quadratic.
Then the overall time constant oI the third-order
autopilot system can be approximated as
2
total
= + (6)
Since the missile perIormance requirements and
hardware limitation, engineers would like to put Iorward
the time- and Irequency-domain mixed perIormance index
in terms oI , , and c
CR
. Wherein, can approximate
reveal system response time, is applies to endow the plant
proper damping, and c
CR
reIers to the crossover Irequency
oI open-loop system.
From the point oI view oI pole assignment, once
Iinding out c, the desired closed-loop system poles can be
determined immediately
2 2
1 1 1 P f f = +
(
For given airIrame parameters and perIormance index,
it can decide the desired poles and the autopilot gains iI
only providing a c value, as well as the open and closed
loop systems. Then c
CR
is determined. In other words, there
exists a bijection between c
CR
and c
( )
CR
f (7)
The Iollowing Iormula can be used to make an
estimate oI c value range
|6|
(1 ) 1 (2 )
appro CR
CR
a b
+
= +
(
(
(8)
Then deIining the objective Iunction with the required
c
CR
and calculated c
cr
every time
min
CR cr
(9)
Hereby an unconstraint optimization problem can be
outlined to obtain the exact c value. The existence and
convergence oI this optimization problem are explicit.
With
m
,
m
and poles , the state Ieedback matrix
can be determined immediately
u x = (10)
Next we will deal with the problem that the Iull state is
not available Ior Ieedback in a missile control problem. By
deIining the "Iull state observability" as requiring that
|
m
m
|
-1
exist, then any state Ieedback optimal solution
can be transIormed to output Ieedback given the system is
Iull state observable. Specially, iI
m
0, the requirement to
transIorm a state Ieedback solution to output Ieedback
reduces to demanding
1
m
ag
reIers to autopilot gains, to two-loop autopilot,
ag
|k
a
k
g
|; to two-loop autopilot with PI compensator,
ag
|k
p
T
i
k
g
|; to three-loop autopilot,
ag
|k
a
c
i
k
g
|.
Besides, one oI the main purposes oI missile control
system is to track the normal acceleration (overload)
command. The real output normal acceleration a
y
is
contained in the Iirst output item (v
1
). With the terminal
value theorem
0
lim 1
v
s
vc
a
a
= (14)
the Iorward-loop gain k
dc
in two- and three-loop topology
can be obtained Irom other gains.
Thereby, the complete autopilot design methodology
can be summarized as: with given mixed index , , and
c
CR
, solving the optimization problem (9) to obtain c value
and the desired poles , checking and determining the
|
m
m
|
-1
existence; Iinding out the state Ieedback matrix
, calculating the output Ieedback matrix ; solving
equation (13) to obtain the autopilot gains. This is an
iterative procedure.
2008 Chinese Control and Decision Conference (CCDC 2008) 3661
It`s turn to reveal how to decide the exact selection oI
desired perIormance index. As shown in Iormer autopilot
topologies, the main hardware comprises: the inertial
measurement units (a rate gyro and an accelerometer), the
Iin applied as plant actuator, and a structural Iilter. The
corresponding hardware parameters are shown in Table 2.
Table2. Hardware Parameters
phase lag (deg)
hardware
damping
Irequency
c (rad/s)
40
(rad/s)
45 (rad/s) 50 (rad/s)
Iin 0.65 220 13.74 15.51 17.31
structural Iilter 0.5 314 7.38 8.32 9.28
rate gyro 0.65 300 10.01 11.28 12.56
accelerometer 0.65 300 10.01 11.28 12.56
The total phase lag led by plant hardware is about
46deg at 45rad/s. iI selecting the open-loop crossover
Irequency as 45rad/s, the Iinal closed-loop system will hold
phase margin no less than 35deg. Thereby, the Ieasible
desired closed-loop system perIormance index can be set as
: 0.2s, closed loop real pole;
0.8, damping ratio oI the closed-loop quadratic
pair;
c
CR
45rad/s, crossover Irequency oI the open-loop
system broken at o.
Wherein, the two-loop topology can only achieve the
index oI 0.8 and c
CR
45rad/s.
These properly-selected time- and Irequency-domain
mixed indexes synthesize requirements oI ultimate agility
in entire Ilight envelope oI the missile and robustness over a
wide range oI mission proIiles at most altitudes.
4 CONTROL ANALYSIS & COMPARISON
4.1 Control Effect on Static Instability
The Iocus oI this paper is to control a static unstable missile
airIrame. With respect to the two-loop topology, Iirstly let
the integral coeIIicient oI PI compensator trend towards
inIinity, i.e., T
i
, k
p
k
a
, the Iinal closed-loop plant will
change into a second-order system same as the common
two-loop topology. It is not diIIicult to obtain the natural
Irequency oI the closed-loop system
CL
( )( )
1 ( )
g p
p
a a b k k J a b a b
k a c Jb
+ + +
=
+
(15)
To inspect the essence, let a
e
0, b
o
0, and c 0
CL
( )
g p
a k k J a b
= + + (16)
It is obvious that the closed-loop system applies body
rate inIormation to increase system damping ratio, and
acceleration inIormation to improve missile airIrame
maneuverability. For the static unstable status, the proper
autopilot gains can stabilize the plant. Thereby, it changes
the plant characteristics by introducing autopilot.
WhereaIter, with respect to the two-loop autopilot
with PI compensator and the three-loop autopilot, as shown
in Figure 6 and Figure 7, both oI the two topologies include
three internal sub-loops, i.e., a damping loop, a stabilizing
loop, and an acceleration (overload) loop. Wherein, the
damping loop increases closed-loop system damping ratio,
the stabilizing loop is to improve system stability, and the
acceleration loop closes the system.
u vc
a
c
Iin
p
k
Acceleration v
a c +
Stabilizing
Damping
g p i
k k c T +
p i
k J T
1
i
p
i
T s
k
T s
+
PI compensator
Fig 6. Equivalent Two-loop Topology
u vc
a
1
dc a i g
k k k
s
Iin
1
a i g
k k
s
Acceleration v
a c +
i g
k
Stabilizing
g
k
Damping
Fig 7. Equivalent Three-loop Topology
What we are interested in is the stabilizing loop when
reIerring to control a static unstable missile. With respect to
the two-loop topology with PI compensator, this sub-loop is
similar to an attitude autopilot with 0
c
and 0 as input
command and output trajectory angle respectively.
Similarly, in the three-loop topology, this sub-loop applies
c
and as the input command and output attitude angle
respectively. Where, both commands are obtained by
integrating autopilot error signal. The stabilizing loop
introduces trajectory angle (two-loop with PI compensator)
or airIrame attitude angle (three-loop) Ior Ieedback, which
approaches to introduce an approximate angle oI attack
Ieedback and equivalent to adjust the distance between CG
and CP. Hereby, both the two- and three-loop topologies
have capacity oI stabilizing a static unstable missile.
4.2 Actuator Analysis
As the internal actuator on missile airIrame, the Iin plays a
crucial role on controlling a missile. During the transient
process, RSS or static instability is useIul during the rising
state as Iorces led by angle oI attack and Iin deIlection
respectively are oI the same direction, the proper static
stability is beneIit to let the plant use least resource Ior
transIerring Irom the rising stage into stable stage, and the
critical static stability makes plant need the least Iin
resource on stable stage. Fin resource dominates the control
Iunction oI autopilot to an unstable missile, however. When
applying RSS technology to decrease the degree oI system
static stability, the Iin will be demanded to provide
increasing control usage Ior holding system stable margin
as the approximate constant. The simulation results in
Figure 8 validate the analysis.
3662 2008 Chinese Control and Decision Conference (CCDC 2008)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
-0.14
-0.12
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
two-loop
two-loop with P
three-loop
a
= -250s
-2
a
= 0
a
= 250s
-2
f
i
n
d
e
f
l
e
c
t
i
o
n
(
d
e
g
)
t(s)
Fig 8. Fin deIlection oI multiple topologies
To holding constant stability margin, the system
requires more actuator resource Ior the lower static
stability. For 50deg phase margin, it plots the required Iin
Irequency vs. a
s
-2
two-loop
two-loop with P
three-loop
Fig 9. Necessary Fin DeIlection oI Two Topologies
Accordingly, though both can stabilize a static
unstable missile, with respect to the unit step response oI
the closed-loop system with a
-250s
-2
, the two-loop
topology with PI compensator is inIerior to the three-loop
topology, as shown in Figure 10. The reason Ior this
phenomenon is actuator resource insuIIiciency on the nose.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
two-loop
two-loop with P
three-loop
a
= -250s
-2
A
c
c
e
l
e
r
a
t
i
o
n
m
/
s
2
t(s)
Fig . Unit Step Response
4.3 Comparison of Two Topologies
Though both with three internal sub-loops, the two-loop
topology with PI compensator is distinct Irom the
three-loop topology.
The essence oI a two-loop topology with PI
compensator approaches a common two-loop topology
with an approximate-cancelling zero-pole pair. While the
three-loop topology approaches a Iirst-order inertial loop.
To comparing two topologies with sub-loops, picking out
the Iirst-order and second-order loops respectively, and
plotting together as shown in Figure 11.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
two-loop with P
second-order loop
three-loop
first-order loop
A
c
c
e
l
e
r
a
t
i
o
n
m
/
s
2
t(s)
Fig 11. Comparison oI Two Topologies
The Iollowing results in Table 3 show the essence oI
two topologies more obviously.
Table3. Closed-loop TransIer Function
a (s
-2
) Two-loop Topology with PI Three-loop Topology
250
2
0.267( 40.46)( 41.96)( 6.353)
( 5)( 38.4 576)
s s s
s s s
+ +
+ + +
2
1.70( 40.46)( 41.96)
( 5)( 38.4 576)
s s
s s s
+
+ + +
0
2
0.36( 43.39)( 44.89)( 4.108)
( 5)( 38.4 576)
s s s
s s s
+ +
+ + +
2
1.48( 43.39)( 44.89)
( 5)( 38.4 576)
s s
s s s
+
+ + +
-250
2
0.43( 46.14)( 47.64)( 3.023)
( 5)( 38.4 576)
s s s
s s s
+ +
+ + +
2
1.31( - 46.14)( 47.64)
( 5)( 38.4 576)
s s
s s s
+
+ + +
Wherein, the actuator dynamics is eliminated to inspect the
problem more clearly. The PI compensator introduces a
zero point to try to cancel the Iirst-order inertial loop in
closed-loop denominator, while the three-loop topology is
dominated by the Iirst-order inertial loop. Thereby, the
2008 Chinese Control and Decision Conference (CCDC 2008) 3663
two-loop topology with PI compensator responds Iaster
than the three-loop topology given there is suIIicient
actuator resource.
5 CONCLUSION
The application oI RSS technology on missile design can
introduce great beneIits to missile maneuverability and
entire Ilight perIormance. The missile with lower static
stability requires more advanced control strategy, however.
Though both oI the two- and three-loop topologies can
stabilize a static unstable missile and achieve to desired
perIormance requirement, the actuator resource is indeed
crucial Iactor aIIecting the autopilot control eIIect.
The PI compensator eliminates system static error and
introduces a zero point to try to counteract the Iirst-order
pole point and make the two-loop topology respond Iaster
than the three-loop topology that has static error ultimately.
The three-loop topology exhibits perIormance similar to a
Iirst-order inertial loop.
REFERENCES
|1| Curtis P. Mracek and D. Brett Ridgely, Missile Longitudinal
Autopilots: Connection between Optimal Control and
Classical Topologies, AIAA 2005-6381.
|2| E. Devaud, H. Siguerdidjane & S. Font, "Some control
strategies Ior a high-angle-oI attack missile autopilot,"
Control Engineering Practice, pp. 885-892, Aug. 2000.
|3| F. W. Nesline and M. D. Nesline, "How Autopilot
Requirements Constrain the Aerodynamic Design oI Homing
Missiles," American Control ConIerence, 716-730, 1984.
|4| Harald Buschek, "Design and Ilight test oI a robust autopilot
Ior the IRIS-T air-to-air missile," Control Engineering
Practice, pp. 551-558, Nov. 2003.
|5| P. Garnell; revised by Qi Zai-kang and Xia Qun-li, Guided
Weapon Control Systems, Beijing Institute oI Technology,
Beijing, 2003.
|6| P. Zarchan, Tactical and strategic missile guidance, 4
th
ed,
AIAA, Washington D C, 2002.
|7| WANG Juan-li and QI Zai-kang, "Analysis oI a three-loop
autopilot," Transactions oI Beijing Institute oI Technology,
Vol.26, No.3, 239-243, 2006.
3664 2008 Chinese Control and Decision Conference (CCDC 2008)