Proceedings of 2012 IEEE

International Conference on Mechatronics and Automation

August 5 - 8, Chengdu, China

Three-loop Autopilot Design and Simulation

CHU Hai-rong ZHANG Yue
Changchun Institute of Optics, Fine Mechanics and Physics, Changchun Institute of Optics, Fine Mechanics and Physics,
Chinese Academy of Sciences Chinese Academy of Sciences
Changchun, Jilin, China Changchun, Jilin, China
[email protected] [email protected]

Abstract – In order to realize missile real-time change flight K R is the gain in the rate
gain in the synthetic stability loop,
trajectory, three-loop autopilot is setting up. The structure
characteristics, autopilot model, and control parameters design loop. KV is the velocity proportional coefficient, K q is the
method were researched. Firstly, this paper introduced the 11th
order three-loop autopilot model. With the principle of systems dynamic pressure proportional coefficient.
reduce model order, the 5th order model was deduced. On that Flight control system closed-loop gain is kept for 1 by
basis, open-loop frequency characteristic and closed-loop K DC , K A KV determines the time constant,
frequency characteristic were analyzed. The variables of velocity
ratio, dynamic pressure ratio and elevator efficiency ratio were K R K q determines the crossover frequency ,and
leading to correct system nonlinear. And then autopilot gains
design method were induced. System flight simulations were WI determines the system damping, respectively.
done, and result shows that autopilot gains played a good job in Because the bandwidth of the gyros, accelerometers, and
the flight trajectory, autopilot satisfied the flight index. elevators are outclass the crossover frequency of three control
loop, there influence can be ignored, and the structure filter
Index Terms – three-loop autopilot , velocity change, dynamic
pressure change.
also can be ignored. Three loop autopilot can be predigest as
figure 2.
a yc a′yc δz K a ( A2 s 2 + A1s + 1)
ay
I. INTRODUCTION K DC KA wi KR
Kq Tm2 s 2 + 2 μmTm s + 1
− KV − s −

Autopilot is an important part of missile. The main G3

K v ( Ti s + 1)
mission is to guarantee flying stability, and make missile Tm2 s 2 + 2 μmTm s + 1
G1
flying with norm trajectory accurately. It is known that the K a ( A22 s 2 + A21 s + 1)
Tm2 s 2 + 2 μ mTm s + 1
U.S. air-to-air guided missile “Sparrow” adopted a lateral
Fig2. the third order Autopilot Model
acceleration autopilot design in the form of three-loop
control[1]. The autopilot is composed of rate loop, synthetic
stability loop, and accelerometer feedback loop. A design III. CONTROLLER DESIGN
method was introduced which autopilot gains did not change
with flight velocity. This method was suitable for the missile The three-loop autopilot with the loop broken right before
which flight velocity changed acutely. the elevators[3-4] as shown in Figure 3.
KA wi KR y x
II. THREE-LOOP AUTOPILOT MODEL KV − s − Kq

G3
Three loop autopilot composed of a rate loop, a synthetic K v (Ti s + 1)
stability loop, and an accelerometer feedback loop[2](Figure 1). Tm2 s 2 + 2 μmTm s + 1
System damping is enhanced by the closed rate loop; the G1
K a ( A22 s 2 + A21s + 1)
absolute stability of the flight control system is governed by −1 Tm2 s 2 + 2 μmTm s + 1
the synthetic stability loop; integral tuning is adopt by the Fig3. Open-loop representation of the three-loop autopilot
accelerometer feedback loop to decrease steady state error.
a yc a′yc δ ac δz ay
K DC
KA wi KR
sF
Ks
•
K a ( A2 s 2 + A1 s + 1) We can write the expression for the open-loop transfer
Gs Ts s + 2μ sTs s + 1 Tm2 s 2 + 2 μmTm s + 1
2 2
KV s Kg
− − − G2
function HG ( s ) as
Kg Kϑ ( Ti s + 1) y K § W W K · K § W W K ·
Gg Tg2 s 2 + 2μ gTg s + 1 G3 Tm2 s 2 + 2 μ mTm s + 1 HG ( s ) = − = − R ¨ −G3 − I G3 − I A G1 ¸ = R ¨ G3 + I G3 + I A G1 ¸
x Kq © s s KV ¹ K q © s s KV ¹

K acc K a ( A22 s 2 + A21 s + 1)

…………………(1)
Ga Ta s + 2 μaTa s + 1
2 2
G1 Tm2 s 2 + 2μmTm s + 1
where G1 and G3 are the airframe aerodynamics transfer
Fig 1. Three-loop Autopilot Model
function, and k ay = kϑV .
There are six control parameter in the three loop After much algebraic manipulation, the open-loop transfer
autopilot. K DC is a closed-loop gain of accelerometer function can be simplify to

instruction, K A is the accelerometer feedback gain, WI is the

978-1-4673-1278-3/12/$31.00 ©2012 IEEE 2499

 T02 s 2 + 2 μ0T0 s + 1 …………(2) From three-loop autopilot open-loop transfer function
HG ( s) = K 0
s (Tmy2 s 2 + 2 μmyTmy s + 1) equation(2), we know HG ( s ) is composed of two
representative quadratic factors and one integral factor.
where the open-loop gain K 0 is govern by According to the principle of frequency response design
method, the phase angle of predigest three-loop autopilot is
KR § K ·
K0 = kϑWI ¨1 + A V ¸ ……………….(3) π § 2μ T ω · § 2μ T ω ·
Kq © KV ¹ ∠HG ( jωcr ) = − − arc tan ¨ my 2my 2cr ¸ + arc tan ¨ 0 20 cr2 ¸
¨ 1− T ω ¸
2 © my cr ¹ © 1 − T0 ωcr ¹
And the coefficients of the numerator polynomials can also be
…...(11)
expressed as
From Figure 2, the close-loop transfer function of three-
§ KA · loop autopilot is expressed as
¨ Tiy + WI VA22 ¸
T02 = ©
KV ¹ …………………...(4) ay ( s ) a y ( s ) δ z1 ( s )
§ KA · = ....................(12)
WI ¨1 + V¸ a yc ( s ) δ z1 ( s ) a yc ( s )
© KV ¹
ay ( s )
§ · Where is transfer function from the missile center of
¨1 + TiyWI + WI
KA
VA21 ¸ δ z1 ( s )
KV
2μ0T0 = © ¹ …………(5)
gravity to the actuatorΔ
§ KA ·
WI ¨ 1 + V¸ ay ( s ) kay ( A2 y s 2 + A1 y s + 1)
© KV ¹ = ……….(13)
The open-loop amplitude ratio is derived
δ z1 ( s ) Tmy2 s 2 + 2 μmyTmy s + 1
δ z1 ( s )
HG ( jω ) =
K0 (1 − T ω ) + ( 2μ T ω )
0
2 2 2
0 0
2
is the transfer function from the actuator to the
a yc ( s )
…..(6)
ω ( −2μ T ω ) + (1 − T ω ) 2 2
2 2
my my my
over loading command,
At the cut-off frequency ωcr , the amplitude ratio is 1. We K A WI K R
obtain the following expression for HG ( jω ) δ z1 ( s ) KV s K q
=
HG ( jωcr ) = 1 ……………….........(7) a yc ( s ) KR § WI W K ·
1+ ¨ G3 + G3 + I A G1 ¸ ...(14)
After much algebraic manipulation, Equations(7) can be Kq © s s KV ¹
rewritten as K A WI K R
m
4 6
cr (
T ω + ( 2 μ myTmy ) − 2T − K T
2 2
my
2 4
0 0 )ω 4
cr
……..(8) =
KV s K q
1 + HG ( s )
( )
+ 1 + 2 K 02T02 − ( 2 μ0T0 ) K 02 ωcr2 − K 02 = 0
2
So
As we know airframe is quadratic factors, the phase angle ay ( s ) a y ( s ) δ z1 ( s )
=
of the airframe is a yc ( s ) δ z1 ( s ) a yc ( s )
­ § 2μ myTmyω ·
° −arc tan ¨¨ 2 ¸
, ω ≤ ωm KA
WI R kay ( A2 y s 2 + A1 y s + 1) K 0
K ...(15)
¸
© 1 − Tmyω ¹
2
°° ….....(9) KV Kq
ϕm (ω ) = ® =
° ª § 2μmyTmyω · º 1 § 2μ my 1 · § 1 2μ0 ·
s3 + ¨ + 2 ¸ s2 + ¨ + ¸ s +1
°− «π − arc tan ¨¨ T 2 ω 2 − 1 ¸¸ » , ω > ωm K 0ωmy
2 ¨K ω ω ¸
© K 0 ω0 ¹
°¯ «¬ © my ¹ ¼» © 0 my 0 ¹

WI R kay ( A2 y s 2 + A1 y s + 1) K 0
And the phase angle of the controllers KA K
KV Kq
T02 s 2 + 2 μ0T0 s + 1 is =
2μ
(1 + τ s ) §¨ 2 s 2 + s + 1·¸
1
­ § 2μ0T0ω · ©ω ω ¹
° arc tan ¨ 2 2 ¸
, ω ≤ ω0
° © 1 − T0 ω ¹ The closed loop poles of three-loop autopilot are
ϕc (ω ) = ® ...(10) typically characterized by a real pole and a quadratic pair of
° ªπ − arc tan § 2μ0T0ω · º , ω > ω KA K
, WI , R can then be chosen to achieve a
°« ¨ 2 2 ¸» poles.
© T0 ω − 1 ¹ ¼
0
¯¬ KV Kq

2500
desired closed real pole(τ ), closed loop damping of the § 1 2·
¨ 2 + τ ¸ + ª¬ 4 μ myωmyτ − 2 + 4 μ (τ ωcr + 1) − 2τ ωmy º¼ 2
2 2 2 2 2 1
quadratic pair( μ ) , and open loop crossover frequency( ωcr ).
© ωcr ¹ ω
The denominators of Equ(15) are equivalent, so
+ 4 μτ ª¬(1 − 2μ myωmyτ ) ωcr2 − ωmy
1
τ 1 2
º 3
¼ω
= ...................................(16-a)
ωcr K 0ωmy
2 2

+ ª«ωcr2 (1 − 2 μ myωmyτ ) + τ 2ωmy

1
(1 − ωcm ) º¼» 4 = 0
2 4
2μ 1 2 μmy 1 ¬ ω
τ+ = + ...........(16-b)
ωcr ω
K 0ωmy ω02
2

( )
cr
where X = 1 − 2 μ myTmyτ .
2μ 1 2μ0
τ+ = + .......................(16-c) Using dichotomy method, newton-downhill method, or
ωcr K 0 ω0
1
2
We can solve K 0 from Equ(7): QR method, can be solved , and ω0 , μ0 , and K 0 can
ω
ωcr2 ª«(1 − Tm2ωcr2 ) + ( 2 μmTmωcr ) º»
2 2 also be solved.
K = 2 ¬ ¼ 2μ 1 2μ myωmyτ
ω0 = 1 τ+ −
( −2μmyTmyωcr ) + (1 − Tmyωcr )
0 2 2 2 2
.........(17) ω ω ω2 2

ωcr2 ωcm 1§ 2 μ τωmy ·

2

= μ0 = ¨¨τ + − 2 ¸ ω0
( −2 μ T ωcr ) + (1 − Tmy2 ωcr2 )
2 2
2© ω ω ¸¹
my my

ω2
ωcm = (1 − Tm2ωcr2 ) + ( 2μmTmωcr )
2 2
Where . K0 =
τωmy
2

We choose airframe damping μmy ,and natural frequency

We have already know the numerator of open-loop
ωmy , control system response parameters τ , μ , ωcr , the transfer function, equation(4) and equation(5) can be rewritten
as
designer chosen gains can be calculated from the following
system of equations. Tiy KA § K ·
+
VA22 = T02 ¨1 + A V ¸ ................(20)
We can now solve for K 0 yielding WI KV © KV ¹
ω2 1 K § K ·
K0 = ……………………(18) + Tiy + A VA21 = 2μ0T0 ¨1 + A V ¸ .........(21)
τωmy
2
WI KV © KV ¹
Substituting K 0 into the second equation(16-b) yielding Equation(20) – Equation(21) h Tiy :
1 2μ 1 2 μmyωmyτ
= τ+ − § K ·
V ( A22 − A21Tiy ) − Tiy2 = (T02 − 2μ0T0Tiy ) ¨1 + A V ¸
KA
ω2
ω ω 2
ω KV
0
The third equation(16-c) yielding
© KV ¹
K
2μ0 τωmy
2
2μ We can solve the preceding equation for A , and WI :
=τ + − 2 KV
ω0 ω ω
Substituting the equation(18) into equation(17) yielding KA T02 − 2 μ0T0Tiy + Tiy2
= ......(22)
ω4 ωcr2 ωcm KV V ( A22 − A21Tiy − T02 + 2 μ0T0Tiy )
=
τ 2ωmy 1 − 2Tmy2 ωcr2 + (Tmy2 ωcr2 ) + 4 ( μmyTmyωcr )
4 2 2
Tiy
WI = .................(23)
...........................(19) § K · K
After some algebraic manipulation, equations(19) can be T ¨1 + A V ¸ − A VA22
0
2

simplified as © KV ¹ KV
On the open loop crossover frequency , flight-control
system open-loop gain is 1,we can solve for the final autopilot
KR
gain yielding
Kq

2501
 KR K0
= .......................(24)
Kq § K ·
kϑWI ¨1 + A V ¸
© KV ¹
V q
Where KV = , K q = , V and q are the average
V q
velocity and dynamic pressure over the entire flight course.
KV and K q are inducted to avoid velocity change.
In addition, we have already derived an expression for (a) K A (b) K R
the close-loop transfer function, it can be predigested as(25)
§ K ·
V ( A2 y s 2 + A1 y s + 1) ¨1 + A V ¸
KA
a y ( s ) KV © KV ¹ .....(25)
=
a yc ( s ) 2μ
(1 + τ s ) §¨ 2 s 2 + s + 1·¸
1
©ω ω ¹
Finally, in order to get unity flight-control system gain
we set the gain of the close-loop transfer function to unity. (d) K DC
And we get
(c) WI
K0 1 Fig 4 autopilot gains
K DC = = 1+ ............(26)
KA K KA
WI R kay V
KV Kq KV
KA
When τ , μ , ωcr are chosen, autopilot gain , WI ,
KV
KR
, K DC and can be computed using Equations
Kq
(a) Elevator angle (b) Flight overload
(22)~(24)ˈand Equation (26).

IV. RESULTS
The autopilot gain algorithm was used on the hypothetical
missile airframe of for the flight condition in which there was
a 1-g step acceleration command when the missile is -25 ̓ .
dynamic characteristic of missile is fine When missile flying
fast, we can increase the cut-off frequency. So when (c) Flight velocity (d) Particle trajectory
V > 480 m s ,we choosing ωcr = 40 rad s , else Fig 5 Flight parameters
ωcr = 35 rad s we also choose τ = 0.1s , μ = 0.75 . Fig Figure 5 shows the missile flight trajectory, elevator angle
4 (a)~(d) shows control parameters over the entire flight is lower than 2.7°. Flight overload command is 1-g when
trajectory which were computed using the autopilot gain trajectory inclination angle lower than -25°, the real flight
algorithm. overload is maintaining 1g, the rise time is about 10 seconds.
The velocity is kept at 200m/s after autopilot responses the
flight overload command. Fig 5(d) also shows the real flight
trajectory. Cut-off frequency is 38.9rad/s, phase margin is 40°,
and magnitude margin is 7dB. Simulation results satisfy the
flight index

2502
 V. CONCLUSIONS
The paper shows the three-loop autopilot design method.
Autopilot gains K A , WI , K R , and K DC can be computed
using this method. Parameters KV and K q are changed with
velocity, they played an important role to gain scheduled. We
ωcr , τ , and μ , autopilot gains were
have chosen flight index
computed. Substituted these gains into flight trajectory, and
simulations were done. Simulation results showed that cut-off
frequency, phase margin, and magnitude margin were 38.9rad/s,
40°, 7dB respectively.

REFERENCES
[1] GARNELL P. Guided Weapon Control Systems. Royal Military College
of Science[M]. Revised by QI Z K and XIA Q L. Beijing: Beijing Institute
of Technology ,2002
[2] F.W Nesline, How autopilot requirements constrain the aerodynamic
design of homing missiles[R], Raytheon Company, Missile Systems
Division.
[3] ZARCHAN P. Tactical and Strategic Missile Guidance[M], American
Institute of Aeronautics and Astronautics. Inc.2002
[4] Katsuhiko Ogata. Modern Control Engineering[M], publishing house of
electronics industry, 2011

2503