Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

4 Analysis of Proportional

Navigation Guided
Systems in the
Time Domain

4.1  INTRODUCTION
It is well known that an investigation of processes and phenomena is
linked, first of all, with the construction of mathematical models describ-
ing these processes and phenomena using mathematical language. The
model is characterized by some parameters. These parameters include
input variables or control actions as they are called, or simply controls,
output variables or output coordinates, or controlled variables, and also
intermediate variables, the so-called state variables. In most cases pro-
cesses are not considered in isolation but in direct connection with other
processes and phenomena. The influence of external conditions—the
environment—is characterized by the so-called disturbing influences or,
simply, disturbances.
As a matter of fact, the mathematical model is nothing but the analytical
expression of an interconnection of the specified parameters. The param-
eters chosen are determined by the problem under consideration.
In Chapter 2, the control theory approach was used to obtain the pro-
portional navigation (PN) guidance law. The line-of-sight (LOS) rate was
considered as the system output; the PN law, the commanded missile
acceleration, was considered as control or input; and the target accelera-
tion was considered as disturbance. The Lyapunov approach, the pivot of
control theory, was used in the previous chapter to obtain a wide class of
guidance laws implementing parallel navigation. Since the PN term is the
main component of the considered guidance laws and, separately, the PN
law produces the lateral motion of unmanned aerial vehicles, we will pay
special attention to this guidance law. Below we will build and analyze the
models of PN guided missile systems. Similar models can be built for the
PN guided UAV systems. The miss distance, the parameter that character-
izes the missile guidance system performance, is the system output. The
missile and target accelerations are control and disturbance, respectively.

65
66 Guidance of Unmanned Aerial Vehicles

For simplicity, we will analyze the planar model with one control action.
The planar model itself is widely used on the preliminary design stage.
As seen from Chapter 3, the results obtained for this model can be easily
generalized for the three-dimensional engagement model.
In control theory, analytical tools were developed for describing the
characteristics of control systems based on the concept of the system error.
The goal of control is to reduce the error to the smallest feasible amount.
The ability to adjust the transient and steady-state response of a control
system to meet certain performance requirements is the main goal of its
design. To analyze systems their performance criterion should be defined.
Downloaded by [University of South Wales] at 05:31 14 October 2013

Then, based on the desired performance, the parameters of the system


or/and its structure should be adjusted to provide the desired response.
Because the actual input signals are usually unknown, a standard test input
signal is normally chosen. The time-domain analysis is usually based on
the so-called step input.
The miss distance in the guidance system analysis and design is, at a
certain degree, analogous to the error in conventional control systems. The
goal of guidance is to reduce the miss distance to the smallest feasible
amount. Target maneuver plays a major role in determining missile sys-
tem performance. The miss distance, due to a step-target maneuver, is the
miss step response similar to the well-known time-domain characteristic
in control theory. Below we obtain analytical expressions of miss distance
for simple models of PN guidance systems. Unfortunately, in the time
domain the closed-form solutions cannot be obtained for high-order mod-
els realistically reflecting autopilot and airframe dynamics. Nevertheless,
the models under consideration enable us to establish some properties of
linear models of PN guided missile systems.

4.2  INERTIALESS PN GUIDANCE SYSTEM


Although PN guidance presents a nonlinear control problem, to apply a
known technique of analysis and design, the system equations are linear-
ized yielding a linear time-varying system. The linearization is valid when
it is assumed that the missile and target approach the so-called collision
course. The results of simulation of linear and nonlinear models show that
the linearized model faithfully represents the guidance system dynamics,
i.e., the linearization is valid close to the interception where the closing
velocity can be considered constant so that the range can be approximated
by a linear function of time [1,2,5].
In Chapter 2 we considered equation (2.35) to obtain the expression for
the PN guidance law. We will use this equation to analyze the performance
of the idealized linearized inertialess model of the PN missile guidance
system (2.35). Substituting equation (2.7) into (2.35) we have:
Analysis of Proportional Navigation Guided Systems in the Time Domain 67

y(t ) = − Nvcl λ (t ) + aT (t ) (4.1)

After integration of both parts of equation (4.1) and taking into account
equations (2.9) and (2.11), it becomes:

N
y (t ) = − Nvcl λ(t ) +
∫ a (t)dt = − t
T
F −t
y(t ) + VT (t ) (4.2)

The solution of equation (4.2) is presented in the following form:


Downloaded by [University of South Wales] at 05:31 14 October 2013

y(t ) =
1
M (t )
( ∫ V (t)M (t)dt + C )
T (4.3)

where C is a constant of integration and

 N
dt  = (t F − t )− N
M (t ) = exp
 ∫t F −t

(4.4)

It can be simplified as:

tF − t (t − t ) N
y(t ) = C (t F − t ) N +
N −1
VT (t ) − F
N −1 ∫ (t F − t )− N +1 aT (t )dt (4.5)

In the case of a step target maneuver, i.e., aT(t) = aT the last expression has
the form:

 (t F − t )t (t F − t )2 
y(t ) = C (t F − t ) N + − a
 N − 1 ( N − 1)( N − 2)  T (4.6)

where the constant C is determined based on the initial conditions


for y(t).
The analysis of equation (4.6) enables us to conclude that the miss
distance y(tF) is zero, i.e., proportional navigation with the effective navi-
gation ratio N > 2 is an effective way to hit a target. To be more rigorous,
we should mention that the expression (4.6) indicates that only values
N = 1 and N = 2 are dangerous. But by choosing N > 2 we guarantee
zero miss.
The model of the missile guidance system considered above is too
simple to make immediate optimistic estimates of the PN law perfor-
mance. Even a slightly more complicated linear model of the missile
68 Guidance of Unmanned Aerial Vehicles

guidance system (e.g., by presenting autopilot inertia by the first-order


dynamic unit) makes the problem of the miss distance analysis very
complicated.
The miss distance model for a missile with a first-order acceleration lag
τ1, against a target undergoing a step acceleration maneuver, is described
by the following equations:

y1 = y2

y 2 = aT − aM
Downloaded by [University of South Wales] at 05:31 14 October 2013

a M = (ac − aM ) /τ1

a T = δ(t ) (4.7)
y(t ) = y1

 y1 y
ac (t ) = N + 2 
 (t F − t ) 2 tF − t 

In this case the commanded acceleration ac(t) does not coincide with the
real missile acceleration aM(t) and is presented in the form (2.14). A step
maneuver aT(t) at t = 0 is described by a differential equation with the
delta-­function δ(t) in its right part.
It is impossible to obtain a visible analytical solution y(t) of the above
linear equation with time-varying coefficients and the singularity at t = tF.
The general approach to analysis of this type of equation is the use of
simulation tools. Since our main interest lies in analyzing the miss dis-
tance y(tF), it means that we should simulate the system (4.7) for various
tF. To avoid multiple simulation trials and obtain y(tF) in one computer
run, the method of adjoints is used [4,5]. Moreover, a specific structure of
equation (4.7) enables us, based on the method of adjoints, to obtain the
analytical solution of equation (4.7) with respect to y(tF).

4.3  METHOD OF ADJOINTS


The method of adjoints, which is a useful tool to simulate the impulse
response P(σ, t) of time-varying linear systems for the fixed observa-
tion time σ = tF with respect to the impulse application time t, has been
widely used in missile guidance system design and analysis, especially
for linearized engagement models. An approach to obtaining the adjoint
system is based on a structural representation of the guidance system
model.
Analysis of Proportional Navigation Guided Systems in the Time Domain 69

The method of adjoint will be explained on the example of a linear


time-­varying system described by the system of the differential equation
presented in the vector-matrix form:

y = A(t F − t ) y + f , 0 ≤ t ≤ tF (4.8)

where y and f are n-dimensional vectors, A is a matrix with coefficients


depending on tF – t.
By introducing the system adjoint to the system (4.8) as

x = − AT (t F − t ) x (4.9)
Downloaded by [University of South Wales] at 05:31 14 October 2013

it is easy to check that the adjoint vector x satisfies the condition:

d (x T y)
= xT f (4.10)
dt
or
tF
x T (t F ) y(t F ) − x T (0) y(0) =
∫ 0
x T (σ ) f (σ )d σ (4.11)

where the upper symbol “T” denotes transposition.


To present the miss distance y(tF) due a constant target maneuver, we
should put xT(tF) = (1, 0,…,0) and f(t) = δ(t), so that:

y(t F ) = x T (0) y(0) (4.12)

It is easy to verify that the transition matrix of the adjoint system (4.9)
Φa(t, t0) = ΦT(t, t0), where Φ(t, t0) is the transition matrix of equation (4.8).
For the class of guidance problems under investigation, we should pres-
ent disturbances (target acceleration and other external factors) as the result
of the solution of a system of differential equations. As seen from equa-
tion (4.7), for the case of a step acceleration maneuver (see the condition
f (t ) = δ(t )) it reduces to a simple operation of differentiation.
The initial conditions x(0) of the adjoint system (4.9) can be obtained by
integrating equation (4.9) backward in time, or by considering the modified
adjoint system with respect to time τ = tF – t, i.e., the miss distance y(tF) can
be obtained in one run by simulating the system:

z = AT (τ) z, y(t F ) = z T (t F ) y(0) (4.13)

where zT(0) = (1, 0,…,0).


70 Guidance of Unmanned Aerial Vehicles

For equation (4.7) the modified adjoint system has the following form:

N
z1 = z3
τ1τ 2
N
z 2 = z1 + z3
τ1τ
(4.14)
1
z3 = − z2 − z3
τ1
Downloaded by [University of South Wales] at 05:31 14 October 2013

z 4 = z2

with the initial condition zT(0) = (1, 0,…,0).


The matrix of coefficients of equation (4.13) is transposed with respect to
the matrix A(tF – t) of the initial system (4.8). Hence, the adjoint modified
system can be modeled by changing inputs by outputs and vice versa in all
elements of the initial system (4.8) and by changing time t in the arguments
of all time-varying coefficients by tF – t. Changing inputs by outputs is
equivalent to the following structural changes: nodes of the original system
become summation units of the modified adjoint system, summation units
of the original system become nodes of the modified adjoint system, and the
direction of all signal flow is reversed. In addition, as mentioned above, the
structural changes of the original system may be needed to convert its actual
input to the equivalent impulsive input.
By differentiating the second equation of the system (4.14), it can be
transformed to:
N
z2 = z3 (4.15)
τ1τ

Using the Laplace transform and substituting z3 from the third equation of
(4.14), equation (4.15) can be presented as:

d 2 Ns 2
(s Z 2 (s)) = Z 2 (s )
ds s(τ1s + 1)
or
d
X (s) = NH (s) X (s) (4.16)
ds

where s is the symbol of the Laplace transform:

X (s ) = s 2 Z 2 (s ) (4.17)
Analysis of Proportional Navigation Guided Systems in the Time Domain 71

and

1 W (s )
H (s ) = = (4.18)
s(τ1s + 1) s

The solution of equation (4.16) can be presented as:

(∫ )
N
 s 
X (s) = s 2 Z 2 (s) = C exp NH (s)ds = C (4.19)
 s + 1/τ1 
Downloaded by [University of South Wales] at 05:31 14 October 2013

where C is a constant determined by the initial conditions.


From the last equation of the system (4.14) we obtain:

N
 s 
Z 4 (s) = s −3C = s −1 X ( s ) (4.20)
 s + 1/τ1 

The constant C = 1 is determined from the condition z1 (0) = z 2 (0) = 1, which
follows from the first two equations of (4.14), so that lim s 2 Z 2 (s) = lim X (s) = 1
s →∞ s →∞
and equation (4.20) becomes:

N
 s 
Z 4 (s ) = s−3
(4.21)
 s + 1/τ1 

Taking into account equations (4.13) and (4.21), for the effective navi-
gation ratio N = 4 the miss distance due to the unit step target accelera-
tion can be presented as:

y(t F ) = t F2 exp(−t F /τ1 )(0.5 − t F / 6τ1 ) (4.22)

The block diagram of the original missile guidance system (4.7) is given in
Figure 4.1 (D denotes the operator of differentiation).
The adjoint system (4.14) structure is presented in Figure 4.2. The
above-given block diagrams of the original (4.7) and adjoint (4.14) systems
for the unit step-target acceleration aT = 1 can be simplified. Their simpli-
fied form, based on equations (2.10)–(2.14), is presented in Figure 4.3 and
Figure 4.4, respectively.
The modified systems are more convenient for the analysis. The modi-
fied original system operates directly with the line-of-sight λ = y/(vclτ) and
its derivative. The modified adjoint system corresponds to the transforma-
tion (4.16)–(4.18); the closing velocity vcl is shown in Figure 4.4 to corre-
spond fully to Figure 4.3.
72 Guidance of Unmanned Aerial Vehicles

y(t) ac aM
aT 1 1 N 1
δ(t) D D2 τ2 τ1D + 1

N
D
τ

FIGURE 4.1  Block diagram of original guidance system.


Downloaded by [University of South Wales] at 05:31 14 October 2013

z4 z2 δ(τ) z3
aT 1 1 N 1
y(τ)
D D2 τ2 τ1D + 1

–1 N
D
τ

FIGURE 4.2  Block diagram of adjoint system.

y λ λ aM
1 1 T y 1 Nvcl
aT D
D D2 – vclτ τ1 D + 1
δ(t)
yM
1
D2

FIGURE 4.3  Modified block diagram of original guidance system.

The result of simulation of the adjoint system shown in Figure 4.5 for
τ1 = 0.5s and aT = 1g (the acceleration of gravity g = 9.81 m/s2) presents the
miss step response of the missile guidance system.
As seen in Figure 4.5, in contrast to the idealized linearized inertialess
model (4.1), the miss distance of the inertial missile guidance system (4.7)
is not zero. The acceleration time lag τ1 significantly influences the miss
step characteristic. It is obvious [see, e.g., equation (4.22)] that the miss
distance is smaller for a smaller τ1.
The analysis of the analytical expressions (4.6) and (4.22) for the miss
distance due to a step target maneuver allows us to conclude that, despite
Analysis of Proportional Navigation Guided Systems in the Time Domain 73

δ(τ)
z4 z2
aT 1 1 1 Nvcl
D D2 vclτ D τ1 D + 1
y(τ)

–1
D2
Downloaded by [University of South Wales] at 05:31 14 October 2013

FIGURE 4.4  Modified block diagram of adjoint system.

1.4

1.2

0.8
Miss (m)

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14 16 18 20
Flight time tF (s)

FIGURE 4.5  Miss distance for step target maneuver.

the commanded acceleration of the linearized PN guidance system model


[see equation (4.7)] tends to infinity when t tends to tF , it does not influ-
ence significantly the miss distance y(tF); it does not influence at all in the
case of the idealized inertialess model.
The miss step response is one of the most used estimates of missile sys-
tem performance. It is an important time-domain characteristic of missile
guidance systems that allows designers to choose appropriate parameters
of the missile guidance system to minimize the miss step. The method
of adjoints was developed to simplify the simulation procedure. However,
the necessity to simulate the system response for each impulse application
time by using the model of the original system or using the adjoint system
to simplify this procedure was stipulated by the inability to obtain an ana-
lytical expression for the miss step that can be used for analysis and design
of missile guidance systems.
74 Guidance of Unmanned Aerial Vehicles

The zero- and single-lag guidance systems are convenient analytical


models; but they did not quite match reality. The binomial representation
1/(1 + sT/n)n, where T is the effective guidance system time constant and
n the system order, used for more accurate high-order guidance system
models in [5], still does not accurately reflect flight control system dynam-
ics. The binomial units are usually used to approximate delay units [3] and,
therefore, they cannot be considered a reliable tool for guidance system
design.
Because of an inability to obtain analytical expressions for the miss step
for the high-order planar models, the simulation process using the method
Downloaded by [University of South Wales] at 05:31 14 October 2013

of adjoints still remains a very useful tool of the time-domain analysis.


The analytical difficulties do not allow researchers to build and analyze
more complicated models that also include the dynamics of a maneuvering
target. The frequency approach to analysis and design of missile guidance
systems, described in the next chapter, enables up to overcome the above-
mentioned difficulties.

REFERENCES
1. Yanushevsky, R., and Boord, W. New Approach to Guidance Law Design,
Journal of Guidance, Control, and Dynamics 28, no. 1 (2005): 162–66.
2. Yanushevsky, R. Concerning Lyapunov-Based Guidance, Journal of
Guidance, Control, and Dynamics 29, no. 2 (2006): 509–11.
3. Yanushevsky, R. Optimal Control of Differential-Difference Systems of
Neutral Type, International Journal of Control, 1 (1989): 1835–50.
4. Zadeh, L., and Desoer, C. Linear System Theory, New York, NY: McGraw
Hill, 1964.
5. Zarchan, P. Tactical and Strategic Missile Guidance, Progress in Astronautics
and Aeronautics. Vol. 176. Washington, DC: American Institute of
Astronautics and Aeronautics, Inc., 1997.

You might also like