Escolar Documentos
Profissional Documentos
Cultura Documentos
ABSTRACT
This report considers the derivation of the mathematical model for a missile autopilot in state space form. The basic equations defining the airframe dynamics are non-linear, however, since the non-linearities are structured (in the sense that the states are of quadratic form) a novel approach of expressing this non-linear dynamics in state space form is given. This should provide a useful way to implement the equations in a computer simulation program and possibly for future application of non-linear analysis and synthesis techniques, particularly for autopilot design of missiles executing high g-manoeuvres. This report also considers a locally linearised state space model that lends itself to better known linear techniques of the modern control theory. A coupled multi-input multi-output (MIMO) model is derived suitable for both the application of the modern control techniques as well as the classical time-domain and frequency domain techniques. This is validated by comparing the model with the other published results, and through both open and closedloop systems simulations. The models developed are useful for further research on precision optimum guidance and control. It is hoped that the model will provide more accurate presentations of missile autopilot dynamics and will be used for adaptive and integrated guidance & control of agile missiles.
Published by DSTO Systems Sciences Laboratory PO Box 1500 Edinburgh South Australia 5111 Australia Telephone: (08) 8259 5555 Fax: (08) 8259 6567 Commonwealth of Australia 2002 AR-012-430 August 2002
Executive Summary
Requirements for next generation guided weapons, particularly with respect to their capability to engage high speed, highly agile targets and achieve precision end-game trajectory, has prompted a revision of the way in which the guidance and autopilot design is undertaken. This report considers the derivation of the mathematical models for a missile autopilot in state space form. The basic equations defining the airframe dynamics are non-linear, however, since the non-linearities are structured (in the sense that the states are of quadratic form) a novel approach of expressing this nonlinear dynamics in state space form is given. This should provide a useful way to implement the equations in a computer simulation program and possibly for future application of non-linear analysis and synthesis techniques. This report also considers a locally linearised state space model that lends itself to better known linear techniques of the modern control theory. A coupled multi-input multi-output (MIMO) model is derived suitable for both the application of the modern control techniques as well as the classical time-domain and frequency domain techniques. This is validated by comparing the model with the other published results, and through both open and closed-loop systems simulations. The models developed are useful for further research on precision optimum guidance and control. It is hoped that the model will provide more accurate presentations of missile auto-pilot dynamics and will be used for adaptive and integrated guidance & control of agile missiles.
Authors
____________________
Thanh Lan Vu
Weapons Systems Division
Thanh Lan Vu joined the Guidance & Control group in the Weapons Systems Dvision in August 1999. She received a Master degree in Computing & Control Engineering from the Institute of Norwegian Technology, Trondheim, Norway, in 1991. She also received a PhD in Mechanical Engineering from the University of Western Australia in 1998. Her thesis presents on-line system identification method and active control of non-linear vibration. Her research interests include: Guidance and Control, Signal Processing, Analog and Digital Filters, Active Control of Vibration, and System Identification. ________________________________________________
____________________
Contents
1. INTRODUCTION ............................................................................................................... 1 2. RIGID BODY DYNAMICS ............................................................................................... 2 2.1 Notation and Convention........................................................................................ 2 2.2 Eulers Equations of Motion ................................................................................... 4 2.3 Linearised model for a two-axis symmetrical airframe ..................................... 9 2.4 Incorporation of accelerometer and gyro measurement model ..................... 12 2.5 Linearised model of the airframe including fin servos ................................... 14 2.6 Lateral Auto-pilot design ...................................................................................... 18 3. VERIFICATION OF THE DEVELOPED MODEL...................................................... 19 4. CONCLUSIONS ................................................................................................................ 23 5. REFERENCES..................................................................................................................... 23 APPENDIX A: STATE EQUATION................................................................................. 25 A.1. Quadratic State Vector .................................................................. 25 A.2. Linearisation of the Quadratic Vector........................................ 26 A.3. Calculation of Inverse Matrix ...................................................... 26 FROM STATE-SPACE FORM INTO TRANSFER-FUNCTION..... 28 MATLAB CODES..................................................................................... 29 C.1. Converting the state-space model into the transfer- functions .................................................................................... 29 C.2. Open-loop and closed-loop simulation .................................... 31
APPENDIX B: APPENDIX C:
APPENDIX D: VALUES OF THE NON-ZERO DERIVATIVES AND PARAMETERS USED IN THE SIMULATION ................................................................ 36
DSTO-TN-0449
1. Introduction
Requirements for next generation guided weapons, particularly with respect to their capability to engage high speed, highly agile targets and achieve precision end-game trajectory, has prompted a revision of the way in which the guidance and autopilot design is undertaken. Integrating the guidance and control function is a synthesis approach that is being pursued as it allows the optimisation of the overall system performance. This approach requires a more complete representation of the airframe dynamics and the guidance system. The use of state space model allows the application of modern control techniques such as the optimal control and parameter estimation techniques to be utilised. In this report we derive the autopilot model that will serve as a basis for an adaptive autopilot design and allow further extension of this to integrated guidance and control system design. Over the years a number of authors [1-3, 6] have considered modelling, analysis and design of autopilots for atmospheric flight vehicles including guided missiles. In the majority of the published work on autopilot analysis and design, locally linearised versions of the model with decoupled airframe dynamics has been considered. This latter simplification arises out of the assumption that the airframe and its mass distribution are symmetrical about the body axes, and that the yaw, pitch and roll motion about the equilibrium state remain small. As a result, most of the autopilot analysis and design techniques, considered in open literature, use classical control approach, such as: single input/single output transfer-functions characterisation of the system dynamics and Bode, Nyquist, root-locus and transient response analysis and synthesis techniques [5,7]. These techniques are valid for a limited set of flight regimes and their extension to cover a wider set of flight regimes and airframe configurations requires autopilot gain and compensation network switching. With the advent of fast processors it is now possible to take a more integrated approach to autopilot design. Modern optimal control techniques allow the designer to consider autopilots with high-order dynamics (large number of states) with multiple inputs/outputs and to synthesise controllers such that the error between the demanded and the achieved output is minimised. Moreover, with real-time mechanisation any changes in the airframe aerodynamics can be identified (parameter estimation) and compensated for by adaptively varying the optimum control gain matrix. This approach should lead to missile systems that are able to execute high gmanoeuvres (required by modern guided weapons), adaptively adjust control parameters (to cater for widely varying flight profiles) as well as account for nonsymmetric airframe and mass distributions. Typically, for a missile autopilot, the input is the demanded control surface deflection and outputs are the achieved airframe (lateral) accelerations and body rates measured about the body axes. Other input/output variables (such as: the flight path angle and angle rate or the body angles) can be derived directly from lateral accelerations and body rates.
DSTO-TN-0449
This report considers the derivation of the mathematical model for a missile autopilot in state space form. The basic equations defining the airframe dynamics are non-linear, however, since the non-linearities are structured (in the sense that the states are of quadratic form) a novel approach of expressing this non-linear dynamics in state space form is given. This should provide a useful way to implement the equations in a computer simulation program and possibly for future application of non-linear analysis and synthesis techniques. Detailed consideration of the quadratic/bilinear type of dynamic systems is given in [4]. This report also considers a locally linearised state space model that lends itself to better known linear techniques of the modern control theory. A coupled multi-input multi-output (MIMO) model is derived suitable for both the application of the modern control techniques as well as the classical time-domain and frequency domain techniques. This is validated by comparing the model with others previously published and through simulation of a decoupled single-input single-output (SISO) system.
T Ixx, Lp Iyy, Mq
Izz, Nr
DSTO-TN-0449
m - mass of a vehicle.
- incidence in the pitch plane. - incidence in the yaw plane. - incidence plane angle. - total incidence, such that: tan = tan cos , and tan = tan sin . T thrust.
Table 2.1: Motion variables Vehicle Body Axes System Angular rates Component of vehicle velocity along each axis Component of aerodynamic forces acting on vehicle along each axis Moments acting on vehicle about each axis Moments of inertia about each axis Products of each inertia Longitudinal and lateral accelerations Euler angles Gravity along each axis Vehicle thrust along the body axis Roll axis X p u X L Ixx Iyz ax gx T Pitch axis Y q v Y M Iyy Izx ay gy Yaw axis Z r w Z N Izz Ixy az gz
1 ( 1 + 2 + 3 + 4 ) , if all four control surfaces are active; or 4 1 1 = ( 1 + 3 ) , or = ( 2 + 4 ) if only two surfaces are active. Positive control 2 2
DSTO-TN-0449
+3
+1
& m( u + wq vr ) = X + T + g x m & m( v + ur wp ) = Y + g y m & m( w uq + vp ) = Z + g x m & & & I xx p ( I yy I zz )qr + I yz ( r 2 q 2 ) I zx ( pq + r ) + I xy ( rp q ) = L & & & I yy q ( I zz I xx )rp + I zx ( p r ) I xy ( qr + p ) + I yz ( pq r ) = M & & & I zz r ( I xx I yy ) pq + I xy ( q p ) I yz ( rp + q ) + I zx ( qr p ) = N .
Here () =
2 2 2 2
DSTO-TN-0449
Equations (2.1 to 2.3) represent the force equations of a generalised rigid body and describe the translational motion of its centre of gravity (c.g) since the origin of the vehicle body axes is assumed to be co-located with the body c.g. Equations (2.3 to 2.6) represent the moment equations of a generalised rigid body and describe the rotational motion about the body axes through its c.g.
Separating the derivative terms and after some algebraic manipulation, Equations (2.1 to 2.3) may be written in a vector form as:
~ ~ u 0 0 0 1 0 1 uq X + T g x ~ d v = 0 1 0 0 1 0 ur + Y + g y dt ~ w 1 0 1 0 0 0 vp Z g z vr wp wq
(2.7)
where: X =
X ~ Y ~ Z ~ T ; Y = ; Z= ; T = . m m m m
Note that the states (u,v,w,p,q,r) appear as quadratic terms (form). Equations (2.4 to 2.6) can be written as:
p2 L p [A] d q = [B ] pq + M dt pr N r 2 q qr 2 r
where: matrices [A] and [B ] are given by:
(2.8)
DSTO-TN-0449
I xx I xy I zx [A] = I xy I yy I yz I zx I yz I zz 0 [B ] = I zx I xy I zx I yz I xy (I zz I xx ) 0 I yz (I xx I yy ) I yz I xy
(I
yy
I zz ) I yz I xy I zx 0 I zx
Here again, the states (p,q,r) appear in quadratic form. Equation (2.8) may also be written as:
p2 p d q = [A]1 [B ] pq + [A]1 dt pr r 2 q qr 2 r
L M N
(2.9)
[A]
I yy I zz I yz 2 (I zz I xy + I yz I zx ) (I yz I xy + I yy I zx ) 1 2 (I xx I yz + I zx I xy ) = (I zz I xy + I yz I zx ) I xx I zz I zx 2 (I yz I xy + I yy I zx ) (I xx I yz + I xy I zx ) I xx I yy I xy
(2.9a)
and = I xx I yy I zz I xx I yz I yy I zx I zz I xy 2 I yz I zx I xy .
The selection of the particular order of the terms in the quadratic-state vectors Equation (2.7) and discussed in Appendix A.1.
[uq ur vp vr wp wq ]T of
[p
pq pr q 2 qr r 2
] of Equation (2.8) is
T
Combining Equations (2.7) and (2.9), we obtain the full 6th order rigid body dynamics state equations as:
DSTO-TN-0449
x [11] [C ] d = dt [1] x [0 ] 2
x [12 ] [I ] + 1 x [2 ] [0 ] [A] [B ] 2
[0 ]
[0 ]
1
[[A] ]
u [11] g + u [1] 0 2
(2.10)
0 0 0 1 0 1 where [C] = 0 1 0 0 1 0 1 0 - 1 0 0 0 x 1 = [u v w] ,
[1 ]
T
x 2 = [p q r ] ,
[1 ]
T
x 1 = [uq ur vp vr wp wq ] ,
[2 ]
T
x2 = p2
[2 ]
pq pr q 2 qr r 2 , ~T Z ,
T
~ ~ ~ [1] u1 = X + T Y u 2 = [L M
[1]
N] , gz
g = gx
gy
].
T
[C ] [F ] = [0 ] [I ] [G ] = [0 ]
[0 ]
[0]
1
[[A] ]
DSTO-TN-0449
x = x1 x
[2 ]
[1 ]
[1 ]
x2
[1 ]
] = [u
T
v w
= x1
[2 ]
x2
[2 ]
]
T
[uq
ur vp vr ~ Y ~ Z
wp wq
p2
pq
pr
q2
qr
r2] :
T
is
u = u1
[1 ]
[1 ]
u1
[1 ]
~ ~ ] = [X + T
L M
= g 0 = [g x
T
gy
gz
Note that for a two-axis symmetrical airframe, Iyz = Izx = Ixy = 0. Hence, in this case, the equation (2.9) can be reduced to:
p 0 d q = 0 dt ~ r I zz
where I xx =
0 ~ I yy 0
~ ~ I xx pq L ~ 0 pr + M ~ 0 qr N
(2.12)
I yy I zz ~ I xx I yy I I xx ~ , I yy = zz , I zz = , I xx I zz I yy
~ L ~ M ~ N L= , M = , N= . I xx I yy I zz
As a result, the state equation for the system now becomes:
[I ] + [H ] x [22 ] [0] ~ I xx 0 , 0
[ [0] x 12 ]
g + [1] [I ] u 2 0
[ [0] u 11]
(2.13)
0 ~ I yy 0
x 2 = [ pq
[2 ]
pr
qr ] , and
T
DSTO-TN-0449
~ [1] u2 = L
Remarks:
~ M
~T N .
Equations (2.11) and (2.13) are complete non-linear description of the full 6-DOF autopilot model. In fact, these equations contain quadratic terms in states and will be classed as the quadratic dynamic model. This type of model is required when autopilot design is undertaken for a missile executing high g- or high angle of attack manoeuvres, and (u, v, w, p, q, r) are not small. A more detailed consideration of the algebraic structure of this type of dynamic systems is given in [4].
~ ~
~ ~ X ~ X ~ Xu = , Xv = , u v ~ ~ X ~ X ~ X = , X = ,
~ ~ ~ ~ X ~ X ~ X ~ X ~ Xw = , Xp = , Xq = , Xr = , w p q r ~ X ~ X = ,
~ ~ ~ ~ ~ ~ Y ~ Y ~ Y ~ Y ~ Y ~ Y ~ Yu = , Yv = , Yw = , Yp = , Yq = , Yr = , u v w p q r ~ ~ ~ ~ Y ~ Y ~ Y Y = , Y = , Y = , ~ ~ ~ ~ ~ ~ Z ~ Z ~ Z ~ Z ~ Z ~ Z ~ Zu = , Zv = , Zw = , Zp = , Zq = , Zr = , u v w p q r ~ ~ ~ Z ~ Z ~ Z ~ Z = , Z = , Z = , ~ ~ ~ ~ ~ ~ L ~ L ~ L ~ L ~ L ~ L ~ Lu = , Lv = , Lw = , Lp = , Lq = , Lr = , u v w p q r ~ ~ ~ L ~ L ~ L ~ L = , L = , L = ,
DSTO-TN-0449
~ ~ ~ ~ ~ ~ M ~ M ~ M ~ M ~ M ~ M ~ Mu = , Mv = , Mw = , Mp = , Mq = , Mr = , u v w p q r ~ ~ ~ M ~ M ~ M ~ M = , M = , M = , ~ ~ ~ ~ ~ ~ N ~ N ~ N ~ N ~ N ~ N ~ Nu = , Nv = , Nw = , Np = , Nq = , Nr = , p q u v w r ~ ~ ~ N ~ N ~ N ~ N = , N = , N = .
The six equations of motion of an airframe (using equation (2.13)) can thus be written as:
& u = r0 v + v0 r q 0 w w0 q ~ ~ ~ ~ ~ ~ ~ ~ ~ + X u u + X v v + X w w + X p p + X q q + X r r + X + X + X ~ + T + g x
( ( (
& v = p 0 w + w0 p r0 u u 0 r ~ ~ ~ ~ ~ ~ ~ ~ ~ + Yu u + Yv v + Yw w + Y p p + Yq q + Yr r + Y + Y + Y + g y
(2.14a)
(2.14b)
& w = q 0 u + u 0 q p0 v v0 p ~ ~ ~ ~ ~ ~ ~ ~ ~ + Z u u + Z v v + Z w w + Z p p + Z q q + Z r r + Z + Z + Z + g z
(2.14c)
~ & p = I xx (q 0 r + r0 q ) ~ ~ ~ ~ ~ ~ ~ ~ ~ + Lu u + Lv v + Lw w + L p p + Lq q + Lr r + L + L + L
)
(2.14d)
~ & q = I yy (r0 p + p 0 r ) ~ ~ ~ ~ ~ ~ ~ ~ ~ + M u u + M v v + M w w + M p p + M q q + M r r + M + M + M
(2.14e)
~ & r = I zz ( p 0 q + q 0 p ) ~ ~ ~ ~ ~ ~ ~ ~ ~ + N u u + N v v + N w w + N p p + N q q + N r r + N + N + N
)
(2.14f)
10
DSTO-TN-0449
~ ~ ~ ~ ~ ( r0 + X v ) ( q0 + X w ) Xp ( w0 + X q ) ( v0 + X r ) u ~ ~ ~ ~ ~ Yv ( p 0 + Yw ) ( w0 + Y p ) Yq ( u 0 + Yr ) v ~ ~ ~ ~ ~ w ( p0 + Z v ) Zw ( v 0 + Z p ) ( u0 + Z q ) Zr ~ ~ ~ ~ ~ ~ ~ + Lv Lw Lp ( I xx r0 + Lq ) ( I xx q0 + Lr ) p ~ ~ ~ ~ ~ ~ ~ Mv Mw ( I yy r0 + M p ) Mq ( I yy p0 + M r ) q ~ ~ ~ ~ ~ ~ ~ Nv Nw ( I zz q0 + N p ) ( I zz p0 + N q ) Nr r ~ ~ T + g x X ~ Y g y ~ Z g z ~ + L 0 ~ M 0 (2.15) ~ N 0
(2.16)
11
DSTO-TN-0449
~ ~ ~ Xp v0 + X r w0 + X q ~ ~ ~ w0 + Y p Yq u 0 + Yr ~ ~ ~ u0 + Z q Zr v0 + Z p ~ ~ ~ ~ ~ , Lp I xx r0 + Lq I xx q 0 + Lr ~ ~ ~ ~ ~ I yy r0 + M p Mq I yy p 0 + M r ~ ~ ~ ~ ~ I zz q 0 + N p I zz p 0 + N q Nr
~ ~ ~ T = T0 + T , g x = g x 0 + g x , g y = g y 0 + g y , g z = g z 0 + g z .
pm = p qm = q rm = r
where pm, qm and rm are the measured body rates. Normally, errors due to drifts and noise are included. These appear as additional additive terms in equations (2.17a) to (2.17c). In contrast to the readings of the angular rate components, the readings of the acceleration components are dependent on the location of the accelerometers, w.r.t. the c.g. of the body.
12
DSTO-TN-0449
The acceleration components measured at point O (where O is at a distance of dx, dy and dz from the central point of gravity, c.g., along x-, y- and z-axis, respectively), may be written as: & & & a x = u + qw rv d x (q 2 + r 2 ) + d y ( pq r ) + d z ( pr + q ) (2.18a)
(2.18b) (2.18c)
If the accelerometers are mounted along the x-axis (ie. dy = dz = 0) which is usually the case, then equations (2.18a-c) reduce to:
Note that the right hand side of Equations (2.19a) to (2.19c) come directly from Equations (2.1) to (2.3). Linearising Equations (2.19a) to (2.19c), and using the relationship (2.15) gives us:
y( t ) = [H 1 ]x( t ) + [J 1 ]u( t ) + v( t )
where:
0 0 0 [H 1 ] = 1 0 0
(2.20)
T
13
DSTO-TN-0449
measurement matrix,
[J 1 ] = ~ Y ~ Z
0 0 0 ~ X ~ + N d x ~ M dx
~ Y ~ Z
0 0 0 ~ X ~ + N d x ~ Mdx
~ Y ~ Z
u 1 ( t ) = [
v1 ( t ) = 0 0 0
T + g x m
g y
ks , = 2 2 s s d s + +1 2 s s
(2.21a)
ks , = 2 2 s s d s + +1 2 s s ks , = 2 2 s s d s + +1 2 s s
(2.21b)
(2.21c)
where d, d and d are the demand aileron, elevator and rudder deflection, respectively. ks, ks, and ks are the servo gain for the aileron, elevator and rudder, respectively.
14
DSTO-TN-0449
s, s, and s are the damping factor for the aileron, elevator and rudder, respectively. s, s, and s are the natural frequency for the aileron, elevator and rudder, respectively.
Equations (2.21a) to (2.21c) can be converted into differential equations as follows:
2 2 & & & = s 2 s s + k s s d ,
(2.22a)
&& & = s 2 s s + k s s d ,
2 2 2 2 & & = s 2 s s & + k s s d .
(2.22b) (2.22c)
Hence, the state-space model for the autopilot of a missile including the servos and airframe is: & (2.23) x 2 ( t ) = [A2 ]x 2 ( t ) + [B2 ]u 2 ( t ) + w2 ( t ) , where
u 2 ( t ) = [ d d d ] ,
T
The inputs are now the demanded aileron, elevator and rudder deflection.
T w2 ( t ) = + g x g y g z 0 0 0 0 0 0 0 0 0 , m
15
DSTO-TN-0449
[A2 ] = 0 0 0 0 0 0
[F ]
1
[G ]
1
0 0 0 0 0 0 0 0 0 0 0 s
2
0 0 0 0 0 0 0 1 0 0 2 s s 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0 2 0 s 0 0 0 0
0 0 0 0 s 0
2
1 0 0 2 s s 0 0
0 0 1 0 0 2 s s
0 0 0 0 0 0
0 0 0 0 0 0 [B2 ] = 0 0 0 2 k s s 0 0
0 0 0 0 0 0 0 0 0 0 k s s 0
2
. 2 k s s
0 0 0 0 0 0 0 0 0 0 0
y 2 ( t ) = H 2 x 2 ( t ) + v 2 ( t ) ,
where
(2.24)
y 2 ( t ) = p m
q m
rm
a x m
a y m
a z m ,
16
DSTO-TN-0449
(Note that gyro drift and noise and the accelerometer bias may be added to the right hand side of Equation (2.24)).
[H 2 ] =
[H ]
1
M M M M M M
[J ]
1
M M M M M M
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 , 0 0 0
v 2 ( t ) = 0 0 0 0 0 0 0 0 0
T + g x m
g y
g z .
Figure 2.3 shows the block diagram of an open-loop autopilot which contains the fin servos and airframe.
Demand input
Fin deflections
Outpu
u 2
Fin servos
Airframe
y 2
Demand input
u 2
B2
x 2
x 2
H2
Output
y 2
A2
Figure 2.3 A block diagram of an open-loop autopilot system.
17
DSTO-TN-0449
References
Fin deflections
Outpu
Airframe
y 2
Referencesu2
B2
x 2
x 2
H2
Output
y 2
A2 K
Figure 2.4 A block diagram of a closed-loop auto-pilot system. Ignoring the instrument (gyro, accelerometer) dynamics, the measured roll, pitch and yaw angular rates (the gyro outputs) can be expressed as inputs to the gyros multiplied the gyro gains, Kgr, Kgp and Kgy, respectively. Similarly, the measured longitudinal acceleration, ax, and lateral accelerations, ay and az, are inputs to the accelerometers multiplied accelerometer gains, Kax, Kay and Kaz, respectively. The accelerometer gains affect the steady state response and may be set to 1 for transient tests. Rescaling accelerometer gains, after selecting gyro gains, allows a unity gain autopilot to be designed.
18
DSTO-TN-0449
The reference signals, generally used for testing the transient time response of the autopilot, are the desired accelerations in yaw direction, ayd, the pitch direction, azd, and roll rate, pd. The reference roll rate is kept at zero to assess the missile dynamics in roll against spurious disturbances. Hence, the reference vector, r, can be written as:
r = [p d
a zd
a yd ] ,
T
For a case of lateral directional control, the control input signal for the fin servos can be derived as follows:
d = a zd K az a z K gq q d = a yd K ay a y K gr r
(2.25a) (2.25b)
For sake of simplicity, pd is set to zero since this case only considers the lateral directional control. As a result, the control input vector can be written as:
u 2 ( t ) = r( t ) Ky 2 ( t ) ,
where K is the feedback matrix as follows:
(2.26)
0 K = 0 0
0 K gq 0
0 0 K gr
0 0 0 0 0 K ay
0 K az . 0
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Lu , L p , L , M q , M w , M , N r , N v , N , X u , X p , X , Yr , Yv , Y , Z w , Z , Z q ,u 0 , d x ,
we obtain the transfer-function between the roll rate and the aileron deflection as:
~ ~ ~ ~ ~ L s + ( Lu + X X u L ) p( s ) = ~ ~ ~ ~ ~ ~ , ( s ) s 2 ( L p + X u )s + ( X u L p X p Lu )
(2.27)
19
DSTO-TN-0449
The transfer-function between the pitch rate and the elevator deflection as:
~ ~ ~ ~ ~ M s M Zw + M wZ q(s) = , ~ ~ ~ ~ ~ ~ ~ (s) s 2 (M q + Z w )s + M q Z w M w Z q M w u0
The transfer-function between the yaw rate and the rudder deflection as:
(2.28)
~ ~ ~ ~ ~ N s + N vY N Yv r( s ) = ~ ~ ~ ~ ~ ~ ~ , ( s ) s 2 ( N r + Yv )s + N v u0 + N r Yv N vYr
(2.29)
The transfer-function between longitudinal acceleration and the aileron deflection as:
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ a x ( s ) ( X u X + X p X L )s X u X L p + X p X Lu = , ~ ~ ~ ~ ~ ~ ( s ) s 2 ( X u + L p )s + X u L p X p Lu
(230)
The transfer-function between lateral acceleration ay and the rudder deflection as:
ay( s )
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ( Y + N dx )s 2 + ( N vY dx N Yv dx + N Yr N r Y )s + N vY u0 N Yv u0 , = ~ ~ ~ ~ ~ ~ ~ (s) s 2 ( N r + Yv )s + N v u0 + N r Yv N vYr
(2.31)
And the transfer-function between lateral acceleration az and the elevator deflection as:
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 2 a z ( s ) ( Z + M dx )s + ( M w Z dx M Z w dx + M Z q M q Z )s + M Z w u0 M w Z u0 = ~ ~ ~ ~ ~ ~ ~ ( s ) s 2 ( M q + Z w )s M w u0 M w Z q + M q Z w
(2.32) Furthermore, if it is assumed that X u = X = Yr = Lu = 0 , the transfer-function between the roll rate and the aileron deflection may be simplified to:
~ L p( s ) = ~ , ( s ) s Lp
The transfer-function between the yaw rate and the rudder deflection as:
(2.33)
20
DSTO-TN-0449
~ ~ ~ ~ ~ N s + N v Y N Yv r( s ) = ~ ~ ~ ~ ~ , ( s ) s 2 ( N r + Yv )s + N v u0 + N r Yv
(2.34)
And the transfer-function between the lateral acceleration in yaw axis, ay, and the rudder deflection measured at the c.g. can be rewritten as:
ay( s )
~ ~ ~ ~ ~ ~ ~ Y s 2 N r Y s + N vY u0 N Yv u0 = 2 ~ ~ ~ ~ ~ . (s) s ( N r + Yv )s + N v u0 + N r Yv
(2.35)
Equations (2.33) to (2.35) are identical to those of the transfer-functions presented in [P.Garnell and D.J.East [Equations (4.6-6), (4.6-8) and (4.6-7)]. The state-space model was used for simulation of open-loop and closed-loop responses for a typical missile, using the same values as those used in [4] (see Appendices C and D). Figures 2.5 and 2.6 show the lateral accelerations of the missile due to a step input to the rudder and elevator, respectively, for an open loop simulation. As can be seen from these figures, there are large steady state errors. However, the steady state errors can be reduced with a feedback loop as can been seen in Figures 2.7 and 2.8. These simulation results are similar to the results presented in [4].
800
700
500
400
300
200
100
0.5
2.5
Figure 2.5 Open loop simulation: Lateral autopilot response to a step demand acceleration
21
DSTO-TN-0449
800
700
500
400
300
200
100
0.5
2.5
Figure 2.6 Open loop simulation: Lateral autopilot response to a step demand acceleration.
60
50
40
30
20
10
0.02
0.04
0.06
0.08
0.12
0.14
0.16
0.18
0.2
Figure 2.7 Closed loop simulation: Lateral autopilot response to a step demand acceleration.
22
DSTO-TN-0449
60
50
40
30
20
10
0.02
0.04
0.06
0.08
0.12
0.14
0.16
0.18
0.2
Figure 2.8 Closed loop simulation: Lateral autopilot response to a step demand acceleration.
4. Conclusions
Both the non-linear and linearised autopilot models have been derived in this report. The state-space model of a missile autopilot was validated by comparing the model with the other published results, and through both open and closed-loop systems simulation. The non-linear dynamics model presented as structural quadratic algebraic system is novel and will be used for developed non-linear control techniques suitable for missile systems high g- manoeuvres and operating of a range of aerodynamics conditions. The models developed in this report are useful for further research on precision optimum guidance and control. It is hoped that the higher order model with motion and inertial coupling will provide more accurate representation of missile autopilot dynamics and should be used for adaptive and integrated guidance and control of agile missiles.
5. References
1. Babister, A.W., Aircraft Dynamic Stability and Response, Pergamon, 1980. 2. Blakelock, J.H., Automatic Control of Aircraft and Missiles, John Wiley & Sons, Inc., 1965. 3. Cook, M.V., Flight Dynamics Principles, Arnold, 1997.
23
DSTO-TN-0449
4. Faruqi, F.A., On the Algebraic Structure of a Class of Nonlinear Dynamical Systems, DSTO report (to be sent for publication). 5. Garnell, P. and East, D.J., Guided Weapon Control Systems, Pergamon Press, 1977. 6. Kuo, B.C., Automatic Control Systems, 6th Edition, Prentice-Hall International Editions, 1991. 7. Pallett, E.H.J., Automatic Flight Control, 3rd Edition, BSP Professional Books, 1987.
24
DSTO-TN-0449
x 1 = [u v w] ,
[1] [1]
T
(A1.1) (A1.2)
x 2 = [p q r ] ,
T
[1]
= x1
[ []
1
x2
[1]
= [u v w
p q r] ,
T
(A1.3)
[2 ] [1]
We shall consider the quadratic-state vector x corresponding to x . The quadraticstate vector will be defined as a vector whose elements are components of a homogeneous quadratic polynomial of these states taken in the same lexicographic order. That is, the quadratic-state vector may be written as:
[2 ]
= u 2 uv uw up uq ur v 2 vw vp vq vr w 2 wp wq wr p 2 pq pr q 2 qr r 2
(A1.4)
There are 21 terms in this quadratic state vector. Note that the dimension of the quadratic state vector is
This type of representation has been used by other authors [3, 4] when describing high order state combinations of dynamical systems. In the rigid body dynamic equations (2.7) and (2.9), coefficients of a number of these terms are zero. For the sake of simplicity (to avoid setting large number elements in the matrices to zero), only those quadratic states that are associated with non-zero terms are retained. That is, the [2 ] quadratic-state vector x and its partitioned form may be written as:
25
DSTO-TN-0449
[2 ]
= x1
5)
[ []
2
x2
[2 ]
[uq
ur vp vr
wp wq
p2
pq
pr
q2
qr r 2 (A1.
x [2 ] = x [2 ] x [2 ] 1 2
0 0 p0 r0 0 0 0 0 0 0 0 0
0 0 0 0 p0 q0 0 0 0 0 0 0
0 0 v0 0 w0 0 2 p0 q0 r0 0 0 0
u0 0 0 0 0 w0 0 p0 0 2 q0 r0 0
0 u0 0 v0 u 0 v 0 w 0 p 0 q p0 r 0 q0 (A2.1) 2 r0
Note that (u0, v0, w0, p0, q0, r0) are local operating states. We shall write this relationship in a compact form as:
[2 ]
= x 1
[2 ]
x 2 = [X 0 ] x
[2 ]
[1]
(A2.2)
A [G ] = F E
F E B D D C
26
DSTO-TN-0449
Its inverse [G ]
is given by:
[G ]1
BC D 2 1 = (CF + DE ) (DF + BE )
27
DSTO-TN-0449
and
& x ( t ) = Ax( t ) + Bu ( t ) , y( t ) = Cx ( t ) + Du ( t ) .
(B.1) (B.2)
and
(B.3) (B.4)
(B.5)
Y ( s ) = [ C( sI A ) 1 B + D ]U ( s ) = H ( s )U ( s ) .
Hence, the transfer-function of the system is:
(B.6)
H(s) =
(B.7)
28
DSTO-TN-0449
The following m.file converts the state-space model into transfer-functions clear all; % Define the deravatives and parameters syms Xu Xp Xxi f; syms Yv Yr Yxi Yzeta f; syms Zw Zq Zeta f; syms Lu Lp Lxi f; syms Mw Mq Meta f; syms Nv Nr Nxi Neta Nzeta f; syms u v w p q r f; syms xi eta zeta xi_dot eta_dot zeta_dot f; syms xi_d eta_d zeta_d f; syms ks ws mus f; syms ax ay az f; syms dx u0 f; syms s; % Define the elements a11=Xu; a12=0; a13=0; a21=0; a22=Yv; a23=0; a31=0; a32=0; a33=Zw; a41=Lu; a42=0; a43=0; a51=0; a52=0; a53=Mw; a61=0; a62=Nv; a63=0; of the matrix A a14=Xp; a15=0; a16=0; a24=0; a25=0; a26=-u0+Yr; a34=0; a35=u0+Zq; a36=0; a44=Lp; a45=0; a46=0; a54=0; a55=Mq; a56=0; a64=0; a65=0; a66=Nr;
% Define the elements of the matrix B b11=Xxi; b12=0; b13=0; b21=Yxi; b22=0; b23=Yzeta; b31=0; b32=Zeta; b33=0; b41=Lxi; b42=0; b43=0; b51=0; b52=Meta; b53=0; b61=Nxi; b62=Neta; b63=Nzeta; % Define the elements of the matrix C c11=0; c12=0; c13=0; c14=1; c15=0; c16=0; c21=0; c22=0; c23=0; c24=0; c25=1; c26=0; c31=0; c32=0; c33=0; c34=0; c35=0; c36=1; c41=Xu; c42=0; c43=0; c44=Xp; c45=0; c46=0; c51=0; c52=Yv+Nv*dx; c53=0; c54=0; c55=0; c56=Yr+Nr*dx; c61=0; c62=0; c63=Zw+Mw*dx; c64=0; c65=Zq-Mq*dx; c66=Yr;
29
DSTO-TN-0449
% Define the elements of the matrix D d11=0; d12=0; d13=0; d21=0; d22=0; d23=0; d31=0; d32=0; d33=0; d41=Xxi; d42=0; d43=0; d51=Yxi+Nxi*dx; d52=0; d53=Yzeta+Nzeta*dx; d61=0; d62=Zeta-Meta*dx; d63=0; A=[a11,a12,a13,a14,a15,a16; a21,a22,a23,a24,a25,a26; a31,a32,a33,a34,a35,a36; a41,a42,a43,a44,a45,a46; a51,a52,a53,a54,a55,a56; a61,a62,a63,a64,a65,a66]; B=[b11,b12,b13; b21,b22,b23; b31,b32,b33; b41,b42,b43; b51,b52,b53; b61,b62,b63];
C=[c11,c12,c13,c14,c15,c16; c21,c22,c23,c24,c25,c26; c31,c32,c33,c34,c35,c36; c41,c42,c43,c44,c45,c46; c51,c52,c53,c54,c55,c56; c61,c62,c63,c64,c65,c66]; D=[d11,d12,d13; d21,d22,d23; d31,d32,d33; d41,d42,d43; d51,d52,d53; d61,d62,d63]; U=[xi;eta;zeta]; S=[s,0,0,0,0,0; 0,s,0,0,0,0; 0,0,s,0,0,0; 0,0,0,s,0,0; 0,0,0,0,s,0; 0,0,0,0,0,s]; % The transfer-function of the system is: % H (s) =
30
DSTO-TN-0449
% Calculate (SI A ) 1 SIAinv= inv(S-A); % Calculate the transfer-function H=C*SIAinv*B +D; % Display the transfer-function between the lateral % acceleration, ay, and the rudder deflection,. H(5,3)
C.1.
%********************************************************* %This main program simulates the response of the missile %for a step input at the elevator and rudder. clear all; % Call the M-file called values.m values; % Call the M-file called init.m init; % Call the M-file called ssmodel.m ssmodel; for t=0:ts:5 REF=[theta_d;az_d;ay_d]; U=REF - K*Y; X_dot_prev=X_dot; X_dot=A*X+B*U; X= X+(X_dot_prev + X_dot)/2*ts; Y=C*X; accy=[accy,Y(5,1)]; accz=[accz,Y(6,1)]; time = [time,t]; ay_d=50; az_d= 50; theta_d = 0; end; plot(time,acc1,time,acc3); xlabel('Time [s]'); ylabel('Lateral acceleration, a_y [m/s^2]'); title('Closed loop response for a step input of 50 m/s^2 at the rudder'); figure; plot(time,acc2,time,acc3); xlabel('Time [s]');
31
DSTO-TN-0449
ylabel('Lateral acceleration, a_z [m/s^2]'); title('Closed loop response for a step input of 50 m/s^2 at the elevator'); %**************** Values.m ******************************* % Set values of the parameters Xu=0.0; Xp=0.0; Xxi=0.0; Xeta=0.0; Xzeta=0.0; Yv=-3; Yp=0; Yr=0; Yxi=0.0; Yeta=0; Yzeta=180; Lu=0.0; Lp=0.0; Lxi=0.0; Mu=0.0; Mv=0.0; Mw=-1.0; Mq=-3.0; Meta=-500.0; Mzeta=0.0; Nv= 1; Np=0.0; Nw=0.0; Nr=-3; Nxi=0.0; Neta=0.0; Nzeta=-500.0; Zw=-3; Zq=Yr; Zeta =-Yzeta; ks=0.0068; ksz=ks mus=0.7; musz=mus; ws=180; dx =0.5; u0=500; % For the case of open-loop simulation, Kgr=Kay=Kgq=Kaz=0 Kgr= 30.75; Kay = 0.825; Kgq=-Kgr; Kaz = Kay; %*********************** init.m ************************* %Initialise the parameters u=0; v=0; w=0; p=0; q=0; r=0; xi=0; eta=0; zeta=0; xi_dot=0; eta_dot=0; zeta_dot=0; X=[u;v;w;p;q;r;xi;eta;zeta;xi_dot;eta_dot;zeta_dot]; X_dot=X; Y=[0;0;0;0;0;0]; i = 1; ay_d=0; az_d=0; theta_d=0; xi_d(i)=0; eta_d(i)=0; zeta_d(i)=0; ts=0.001; accy=0; accz=0; time=0;
32
DSTO-TN-0449
% ********************* ssmodel.m *********************** % Define the state-space model a1_1=Xu;a1_2=0;a1_3=0;a1_4=Xp;a1_5=0;a1_6=0; a1_7=Xxi;a1_8=0;a1_9=0;a1_10=0;a1_11=0;a1_12=0; a2_1=0;a2_2=Yv;a2_3=0;a2_4=0;a2_5=0;a2_6=-u0+Yr; a2_7=Yxi;a2_8=0;a2_9=Yzeta;a2_10=0;a2_11=0;a2_12=0 a3_1=0;a3_2=0;a3_3=Zw;a3_4=0;a3_5=u0+Zq;a3_6=0; a3_7=0;a3_8=Zeta;a3_9=0;a3_10=0;a3_11=0;a3_12=0; a4_1=Lu;a4_2=0;a4_3=0;a4_4=Lp;a4_5=0;a4_6=0; a4_7=Lxi;a4_8=0;a4_9=0;a4_10=0;a4_11=0;a4_12=0; a5_1=0;a5_2=0;a5_3=Mw;a5_4=0;a5_5=Mq;a5_6=0; a5_7=0;a5_8=Meta;a5_9=0;a5_10=0;a5_11=0;a5_12=0; a6_1=0;a6_2=Nv;a6_3=0;a6_4=0;a6_5=0;a6_6=Nr; a6_7=Nxi;a6_8=Neta;a6_9=Nzeta;a6_10=0;a6_11=0;a6_12=0; a7_1=0;a7_2=0;a7_3=0;a7_4=0;a7_5=0;a7_6=0; a7_7=0;a7_8=0;a7_9=0;a7_10=1;a7_11=0;a7_12=0; a8_1=0;a8_2=0;a8_3=0;a8_4=0;a8_5=0;a8_6=0; a8_7=0;a8_8=0;a8_9=0;a8_10=0;a8_11=1;a8_12=0; a9_1=0;a9_2=0;a9_3=0;a9_4=0;a9_5=0;a9_6=0; a9_7=0;a9_8=0;a9_9=0;a9_10=0;a9_11=0;a9_12=1; a10_1=0;a10_2=0;a10_3=0;a10_4=0;a10_5=0;a10_6=0; a10_7=-ws^2;a10_8=0;a10_9=0;a10_10=-2*mus*ws;a10_11=0; a10_12=0; a11_1=0;a11_2=0;a11_3=0;a11_4=0;a11_5=0;a11_6=0; a11_7=0;a11_8=-ws^2;a11_9=0;a11_10=0;a11_11=-2*musz*ws; a11_12=0; a12_1=0;a12_2=0;a12_3=0;a12_4=0;a12_5=0;a12_6=0; a12_7=0;a12_8=0;a12_9=-ws^2;a12_10=0;a12_11=0; a12_12=-2*mus*ws; b1_1=0;b1_2=0;b1_3=0; b2_1=0;b2_2=0;b2_3=0; b3_1=0;b3_2=0;b3_3=0; b4_1=0;b4_2=0;b4_3=0; b5_1=0;b5_2=0;b5_3=0; b6_1=0;b6_2=0;b6_3=0; b7_1=0;b7_2=0;b7_3=0; b8_1=0;b8_2=0;b8_3=0; b9_1=0;b9_2=0;b9_3=0;
33
DSTO-TN-0449
b10_1=-ks*ws^2;b10_2=0;b10_3=0; b11_1=0;b11_2=-ksz*ws^2;b11_3=0; b12_1=0;b12_2=0;b12_3=-ks*ws^2; c1_1=0;c1_2=0;c1_3=0;c1_4=1;c1_5=0;c1_6=0; c1_7=0;c1_8=0;c1_9=0;c1_10=0;c1_11=0;c1_12=0; c2_1=0;c2_2=0;c2_3=0;c2_4=0;c2_5=1;c2_6=0; c2_7=0;c2_8=0;c2_9=0;c2_10=0;c2_11=0;c2_12=0; c3_1=0;c3_2=0;c3_3=0;c3_4=0;c3_5=0;c3_6=1; c3_7=0;c3_8=0;c3_9=0;c3_10=0;c3_11=0;c3_12=0; c4_1=Xu;c4_2=0;c4_3=0;c4_4=Xp;c4_5=0;c4_6=0; c4_7=Xxi;c4_8=Xeta;c4_9=Xzeta;c4_10=0;c4_11=0;c4_12=0; c5_1=0;c5_2=Yv+Nv*dx;c5_3=0;c5_4=0;c5_5=0;c5_6=Yr+Nr*dx; c5_7=Yxi+Nxi*dx;c5_8=Yeta+Neta*dx;c5_9=Yzeta+Nzeta*dx; c5_10=0;c5_11=0;c5_12=0; c6_1=0;c6_2=0;c6_3=Zw-Mw*dx;c6_4=0;c6_5=Zq-Mq*dx;c6_6=0; c6_7=0;c6_8=Zeta-Meta*dx;c6_9=0;c6_10=0;c6_11=0;c6_12=0;
A=[a1_1,a1_2,a1_3,a1_4,a1_5,a1_6,a1_7,a1_8,a1_9,a1_10,a1_11,a1_12; a2_1,a2_2,a2_3,a2_4,a2_5,a2_6,a2_7,a2_8,a2_9,a2_10,a2_11,a2_12; a3_1,a3_2,a3_3,a3_4,a3_5,a3_6,a3_7,a3_8,a3_9,a3_10,a3_11,a3_12; a4_1,a4_2,a4_3,a4_4,a4_5,a4_6,a4_7,a4_8,a4_9,a4_10,a4_11,a4_12; a5_1,a5_2,a5_3,a5_4,a5_5,a5_6,a5_7,a5_8,a5_9,a5_10,a5_11,a5_12; a6_1,a6_2,a6_3,a6_4,a6_5,a6_6,a6_7,a6_8,a6_9,a6_10,a6_11,a6_12; a7_1,a7_2,a7_3,a7_4,a7_5,a7_6,a7_7,a7_8,a7_9,a7_10,a7_11,a7_12; a8_1,a8_2,a8_3,a8_4,a8_5,a8_6,a8_7,a8_8,a8_9,a8_10,a8_11,a8_12;
34
DSTO-TN-0449
35
DSTO-TN-0449
Appendix D: Values of the Non-Zero Derivatives and Parameters used in the simulation
~ ~ ~ ~ ~ Yv = 3 , Y = 180 , N v = 1, N r = 3, N = 500 , ~ ~ ~ ~ ~ Z w = 3, Z = 180 , M w = 1, M q = 3, , M = 500 , k s = 0.0068 , k s = 0.0068 , k s = 0.0068 ,
36
DSTO-TN-0449
DISTRIBUTION LIST Mathematical Models for a Missile Autopilot Design Farhan A. Faruqi and Thanh Lan Vu
AUSTRALIA DEFENCE ORGANISATION S&T Program Chief Defence Scientist FAS Science Policy shared copy AS Science Corporate Management Director General Science Policy Development Counsellor Defence Science, London (Doc Data Sheet) Counsellor Defence Science, Washington (Doc Data Sheet) Scientific Adviser to MRDC Thailand (Doc Data Sheet ) Scientific Adviser Joint (Doc Data Sheet and distribution list only) Navy Scientific Adviser (Doc Data Sheet and distribution list only) Scientific Adviser - Army (Doc Data Sheet and distribution list only) Air Force Scientific Adviser Director Trials Systems Sciences Laboratory Chief of Weapon Systems Division RLAWS RLMWS RLLWS RLEWT Guidance and Control Group, WSD (6 Copies) Farhan A. Faruqi Thanh Lan Vu DSTO Library and Archives Library Edinburgh 2 copies Australian Archives Capability Systems Staff Director General Maritime Development (Doc Data Sheet only) Director General Aerospace Development (Doc Data Sheet only) Knowledge Staff Director General Command, Control, Communications and Computers (DGC4) (Doc Data Sheet only)
DSTO-TN-0449
Army ABCA National Standardisation Officer, Land Warfare Development Sector, Puckapunyal (4 copies) Intelligence Program DGSTA Defence Intelligence Organisation Manager, Information Centre, Defence Intelligence Organisation Defence Libraries Library Manager, DLS-Canberra Library Manager, DLS - Sydney West (Doc Data Sheet Only) UNIVERSITIES AND COLLEGES Australian Defence Force Academy Library Head of Aerospace and Mechanical Engineering Hargrave Library, Monash University (Doc Data Sheet only) Librarian, Flinders University OTHER ORGANISATIONS National Library of Australia NASA (Canberra) AusInfo OUTSIDE AUSTRALIA INTERNATIONAL DEFENCE INFORMATION CENTRES US Defense Technical Information Center, 2 copies UK Defence Research Information Centre, 2 copies Canada Defence Scientific Information Service, 1 copy NZ Defence Information Centre, 1 copy ABSTRACTING AND INFORMATION ORGANISATIONS Library, Chemical Abstracts Reference Service Engineering Societies Library, US Materials Information, Cambridge Scientific Abstracts, US Documents Librarian, The Center for Research Libraries, US INFORMATION EXCHANGE AGREEMENT PARTNERS Acquisitions Unit, Science Reference and Information Service, UK Library - Exchange Desk, National Institute of Standards and Technology, US National Aerospace Laboratory, Japan National Aerospace Laboratory, Netherlands SPARES (5 copies) Total number of copies: 51
Page classification: UNCLASSIFIED DEFENCE SCIENCE AND TECHNOLOGY ORGANISATION DOCUMENT CONTROL DATA
2. TITLE 1. PRIVACY MARKING/CAVEAT (OF DOCUMENT)
3. SECURITY CLASSIFICATION (FOR UNCLASSIFIED REPORTS THAT ARE LIMITED RELEASE USE (L) NEXT TO DOCUMENT CLASSIFICATION)
Systems Sciences Laboratory PO Box 1500 Edinburgh South Australia 5111 Australia
6b. AR NUMBER 6c. TYPE OF REPORT 7. DOCUMENT DATE
DSTO-TN-0449
8. FILE NUMBER
AR- 012-430
9. TASK NUMBER
Technical Note
10. TASK SPONSOR 11. NO. OF PAGES
August, 2002
12. NO. OF REFERENCES
J9505-21-207
13. URL on the World Wide Web
01/219
DST
31
14. RELEASE AUTHORITY
http://www.dsto.defence.gov.au/corporate/reports/DSTO-TN-0449.pdf
15. SECONDARY RELEASE STATEMENT OF THIS DOCUMENT
No Limitations
17. CITATION IN OTHER DOCUMENTS 18. DEFTEST DESCRIPTORS Yes
This report considers the derivation of the mathematical model for a missile autopilot in state space form. The basic equations defining the airframe dynamics are non-linear, however, since the non-linearities are structured (in the sense that the states are of quadratic form) a novel approach of expressing this non-linear dynamics in state space form is given. This should provide a useful way to implement the equations in a computer simulation program and possibly for future application of non-linear analysis and synthesis techniques, particularly for autopilot design of missiles executing high g-manoeuvres. This report also considers a locally linearised state space model that lends itself to better known linear techniques of the modern control theory. A coupled multi-input multi-output (MIMO) model is derived suitable for both the application of the modern control techniques as well as the classical time-domain and frequency domain techniques. This is validated by comparing the model with the other published results, and through both open and closed-loop systems simulations. The models developed are useful for further research on precision optimum guidance and control. It is hoped that the model will provide more accurate presentations of missile autopilot dynamics and will be used for adaptive and integrated guidance & control of agile missiles. Page classification: UNCLASSIFIED