You are on page 1of 8

Asian Journal of Control, Vol. 2, No. 4, pp.

243-250, December 2000 243

GUIDANCE LAW EVALUATION FOR MISSILE


GUIDANCE SYSTEMS
Chih-Min Lin, Chun-Fei Hsu, Shing-Kuo Chang
and Rong-Jong Wai

ABSTRACT

In missile guidance system, to reduce the interception “miss distance,”


it is important to choose a suitable guidance law and navigation constant. This
paper investigates and compares the system behavior of guidance laws under
different navigation constants. Based on use of the adjoint technique, miss
distance sensitivity analyses which consider the system noise, target step
maneuver, initial heading error and system parameters for different guidance
laws and navigation constants are presented. Based on these analyses, some
suggestions for choosing a suitable guidance law and navigation constant are
given for the design of missile guidance systems. Also, a suggestion for the
optimal escape time for pilots of fighter planes is given.

KeyWords: Guidance law, navigation constant, adjoint system.

I. INTRODUCTION been accepted as best for many applications in missile


systems. Comparison of the miss distance under different
From the perspective of a control system, guidance guidance laws has been reported in [2,3] based on the use
is a matter of finding the appropriate compensation net- of “Monte Carlo” techniques. However, with the “Monte
work to be placed in series with the plant in order to Carlo” technique, a large number of repeated simulations
accomplish an intercept. In order for the pursuer to impact for many cases must be processed and averaged. The
a maneuverable target with a small miss distance, guid- adjoint technique [10,11] is widely used in guidance
ance uses the principles of feedback control. The purpose system design. It is based upon the impulse response of the
of the guidance system is to determine appropriate pursuer system. This technique not only provides the total root
flight path dynamics such that some pursuer objective can mean square (RMS) miss distance, but also shows how
be achieved efficiently. In many applications, the guid- each of the distance terms contributes to the total RMS
ance system is designed so that it makes use of an inertially miss distance. The adjoint technique is capable of analyz-
stabilized tracker (for example, a seeker) that directly ing linear, time-varying noise-driven systems such as a
measures the angular rates between the pursuer and its guidance loop in one computer run. The performance
target in a fixed coordinate frame. For designing homing outputs of above-mentioned at any instant can be esti-
missile guidance systems, a computerized method of mated using this method. Additionally, information that
analysis that relates the effect of statistical disturbances shows the contribution of all the inputs to the total perfor-
(i.e., the measurement noise, random target maneuver mance outputs can be extracted. In [10,11], the adjoint
etc.) to the overall performance of the system is needed. technique was applied in sensitivity analysis of different
To reduce the interception miss distance, the design of the inputs (the target maneuver, disturbance and noise);
guidance law with a suitable navigation constant is however, it has not so far been used in heading error
important. Previons research on guidance laws has been analysis. Meanwhile, performance evaluation of different
reported in the literature [1-9]. Among the proposed guidance laws and navigation constants as well as the
guidance laws, proportional navigation (PN) [2] and aug- effects of system parameters in the seeker model, autopilot
mented proportional navigation (APN) [3] guidance laws, model and noise filter model were not discussed in [10,
due to their simplicity of onboard implementation, have 11]. This paper investigates the problem of comparing the
system behavior of guidance laws, such as the propor-
Manuscript received October 19, 1999; revised February 29,
tional navigation (PN) guidance law [2], augment PN
2000; accepted July 20, 2000. (APN) guidance law [3] and PID guidance law [12], with
Chih-Min Lin, Chun-Fei Hsu, Shing-Kuo Chang and Rong- different navigation constants. The effects of system
Jong Wai are with Department of Electrical Engineering, Yuan- parameters in the seeker model, autopilot model, and noise
Ze University, Chung-Li, Tao-Yuan, 320, Taiwan. filter model are also discussed. Based on use of the adjoint
244 Asian Journal of Control, Vol. 2, No. 4, December 2000

technique, miss distance sensitivity analyses which con- time-varying system driven by white noise is given by
sider the system noise, target step maneuver and initial
heading error for different guidance laws, navigation tf
constants and system parameters are given. Based on σ out(t f) = Φin h 2(t f , t I)dt I , (3)
0
these analyses, some suggestions for choosing a suitable
guidance law and navigation constant are given to reduce
the interception miss distance. Based on the analysis of where Φin is the spectral density of the white noise (as-
target maneuvers, a suggestion for the optimal escape time sumed to be double-sided and stationary) in units of Hz,
for pilots of fighter planes is also given. and σout is the RMS value of the output. As discussed
previously, simulation of Eq. (3) is impractical because of
II. GUIDANCE LAW ANALYSIS BY the many computer runs needed to generate h(t f, t I).
ADJOINT TECHNOLOGY However, by invoking Eq. (2), we find that

tf
In this section, the adjoint technique is briefly
σ out(t f) = Φin h *(t f – t I, 0)2dt I
reviewed. For every linear dynamic system, whether it is 0
deterministic or stochastic, there exists an adjoint system
constructed from the original system (given in block 0
diagram form) by applying the following rules [10,11]: = – Φin – h *(τ , 0)2d τ
tf

(1) Replace t with (t f – t) in the arguments of all the


variable coefficients, where tf is the final time. If the tf
= Φin h *(τ , 0)2d τ . (4)
coefficients are time-invariant, do not change them. 0
(2) Reverse all the signal flows in the block diagram,
redefine branch points as sum points, and vice versa. Therefore, the RMS value of the output of the origi-
This will make the inputs of the original system appear nal system based on a white noise input can be found by
as outputs of the adjoint system and vice versa (i.e., squaring, integrating, and then taking the square root of the
outputs of the original system will appear as initial impulse response of the adjoint system in only one com-
conditions of the adjoint system). puter run.
Using the adjoint system together with the outputs
It can be shown that the impulse response of the described previously, it is possible to plot the RMS miss
adjoint system, h*, and the impulse response of the original distance at interception versus range at the beginning of
system, h, are related by homing in each computer run. Because the RMS miss
distance due to each of the input quantities can be ob-
h*(tf – tI, tf – t0) = h(t0, tI), (1) tained, it is also possible to isolate the factors that have the
greatest influence on RMS miss distance. Using the
where tI is the impulse application time and t 0 is the adjoint system technique is, therefore, much less time
observation time. The importance of this relationship consuming than using the Monte Carlo technique. The
manifests itself when we desire to observe the value of the adjoint technique is well suited to the problem of statistical
impulse response function of the original system at the analysis and optimization of linear-time varying systems.
final time, tf, based on various impulse application times. Consider the following general linear dynamic mis-
The original system would have to be rerun for each sile guidance system [10]:
impulse application time in order to generate h(tf, tI).
However, the adjoint system has to be run only once in
order to generate the equivalent impulse response by
setting the observation time equal to the final time in – 2V 0 0 0 0 0
Eq. (1), yielding
nT 1 0 0 0 0 –1 nT
*
h (tf – tI, 0) = h(tf, tI). (2) y y
y 0 1 0 0 0 0 y
= ⋅
The adjoint impulse response is identical to the D w1 D
0 0 – w1 0 0
impulse response of the original system in every way λ V c(T – t) λ
except that it is generated in a backward manner. w ⋅w nL
The most important feature of the impulse response nL 0 0 1 2 – w 1w 2 – w 2 0
V c(T – t)
of the adjoint system is that it can be used to statistically
analyze the original system in the presence of stochastic 0 0 0 0 NV cw 3 –w 3
inputs. The RMS response at the terminal time of a linear
C.-M. Lin et al.: Guidance Law Evaluation for Missile Guidance Systems 245

range-independent (fading) noise. It then serves as an


2V 0 0 input to the seeker dynamics to produce a measured LOS
rate. The measured LOS rate is then passed through the
0 0 0 noise filter to obtain the estimated LOS rate λ . The
guidance law generates terminal steering commands, nc, in
0 0 0 Us terms of λ , the closing velocity Vc, the navigation constant
+ ⋅ U sn (5) N and the target lateral acceleration nT. Then, the steering
w1
0 w1 U fn command is sent to the autopilot. The target is assumed to
V c(T – t)
w1 ⋅ w2 have constant velocity, and its lateral acceleration is mod-
0 w w eled as the output of a low-pass filter driven by white noise.
V c(T – t) 1 2
These models have the same autocorrelation function as
0 0 0 does a jinking maneuver (a constant magnitude maneuver
with random sign switching). It is assumed that the seeker
measurement of the LOS rate is corrupted by white glint
The performance of three guidance laws, PN, APN noise and white fading noise with spectral densities Φsn
and PID, is compared below. PN guidance law: and Φfn, respectively. The miss distance, Y(tf), is defined
as the relative missile-to-target displacement at the point
nc = N ⋅ Vc ⋅ λ . (6) of closest approach. The nominal values of all the system
parameters used are tabulated in Table 1.
APN guidance law: Using the “Simulink” software program, the adjoint
model of the linear kinematic guidance system was con-
n c = N ⋅ V c ⋅ λ + N ⋅ N T. (7) structed as shown in Fig. 2. In the process of constructing
2 the adjoint model, the three inputs of the original system,
the target maneuver, glint noise and fading noise, all
PID guidance law: become outputs. That is, the outputs in the adjoint system
are now miss distance sensitivities due to the target
n c = N1 ⋅ V c ⋅ λ + N2 ⋅ V c ⋅ λ + N3 ⋅ V c ⋅ λ. (8) maneuver, glint noise and fading noise, respectively. Since
the sensitivities of the adjoint system do not depend upon
The power of the adjoint technique can be demon- the spectral density levels of the disturbance sources, the
strated vividly by considering a classical missile intercep- program does not have to be rerun if the spectral density
tion problem as shown in Fig. 1. A triple time-constant levels change. Superposition permits calculation of the
guidance system is assumed (one lag for the seeker total RMS miss distance as follows:
dynamics, one lag for the noise filter, and one lag for the
autopilot-airframe dynamics). Figure 1 depicts a miss
distance analysis model composed of elements represent- σ y2(t f) Target σ y2(t f) Glint
ing the intercept kinematics plus elements affecting the σ y(t f) = ⋅ Φs + ⋅ Φss +
Φs Φss
intercept guidance dynamics, and illustrates the interde-
pendence between system elements. Based on the line-of-
1/2
sight (LOS) rate, the guidance law generates a lateral
acceleration for the pursuer while trying to bring the σ (t f) Fading
2
y
+ ⋅ Φfn . (9)
projected miss to zero. In the model, a relative position Y s Φfn
is added to a glint noise, the result of which is multiplied
by the inverse of the range in order to produce a true LOS
angle corrupted by glint noise in inertial frame. A mea-
sured LOS angle is then obtained by adding the result to a Table 1. Nominal values of all system parameters.

Glint noise Fading noise


system parameters nominal value
Autopilot-airframe bandwidth w3 10 rad/sec
Target nT .
LOS angle λ Noise filter bandwidth w2 10 rad/sec
1 Y —
— 1 Y Seeker
maneuver s s calculation model Seeker bandwidth w1 20 rad/sec
model function
Target maneuver bandwidth 2V 0.2 sec–1
RMS target acceleration B 161 ft/sec2
nc λ
.
Noise Average number of crossing zeros τ 1 Hz
Autopilot
model
Guidance law filter
mode;
Spectral density of glint noise Φsn 4 ft2/Hz
Spectral density of fading noise Φfn 1 × 10–6 rad2/Hz
Fig. 1. Missile guidance system. Closing velocity Vc 3000 ft/sec
246 Asian Journal of Control, Vol. 2, No. 4, December 2000

III. SIMULATION RESULTS FOR MISSILE presented. The adjoint technique was applied to perform
GUIDANCE SYSTEMS miss distance analyses in only one run simulation.

In this section, some comparisons among different 3.1 RMS miss distance for different guidance laws and
guidance laws with different navigation constants are navigation constants

Target maneuver Glint noise Fading noise


Based on the simulation results, the RMS levels of
sensitivity sensitivity sensitivity the individual miss distance along with the total RMS miss
distance are arranged according to order vs. adjoint time
( )2 ( )2 ( )2 in Table 2, Table 3 and Table 4. (The adjoint time can be
interpreted as either the time-of-flight or time-to-go at
1 1 1
which statistical disturbance occurs). Comparing these

s —
s —
s three guidance laws, it is found that the minimal total RMS
. miss distance error, 4.7530 ft, for the APN guidance
LOS angle λ
1 Y —
— 1 Y Seeker system with N = 2 is better than those of the PN and PID
s s calculation
model
function guidance systems. Comparisons based on the simulations
are given in Fig. 3, Fig. 4 and Fig. 5. Meanwhile, these
simulation results also provide information about the
nc
.
λ Noise stability behavior of the guidance system. It is obvious
Autopilot filter
model
Guidance law that the PID guidance law has wider bandwidth. However,
model
the wider the system bandwidthis, the more the noise
affects the system. A trade-off must be employed in
Fig. 2. Adjoint system of a missile guidance system.

Table 2. RMS miss distance for the PN guidance system.

RMS Miss N=1 N=2 N=3 N=4 N=5 N=6


Distance (ft)
Glint small small 5.7540 7.4891 9.4508 11.7429
Target divergence divergence 4.3240 3.0248 2.8200 2.9402
Fading small small 3.0829 4.6349 6.7539 9.6282
Total divergence divergence #7.8301 9.3123 11.9534 15.4674

Table 3. RMS miss distance for the APN guidance system.

RMS Miss N=1 N=2 N=3 N=4 N=5 N=6


Distance (ft)
Glint small 4.1323 5.7540 7.4891 9.4508 11.7429
Target divergence 1.2911 2.3187 3.0247 4.0573 5.4876
Fading small 1.9616 3.0829 4.6349 6.7539 9.6282
Total divergence #4.7530 6.9274 9.3123 12.3042 16.1465

Table 4. RMS miss distance for the PID guidance system.

RMS Miss N1 = 3 N1 = 3 N1 = 4 N1 = 4 N1 = 3 N1 = 2
Distance (ft) N2 = 1 N2 = 2 N2 = 1 N2 = 2 N2 = 3 N2 = 1
N3 = 0 N3 = 0 N3 = 0 N3 = 0 N3 = 0 N3 = 0
Glint 5.9989 6.2730 7.7467 8.0305 6.5827 4.4000
Target 5.4021 6.4275 3.5798 4.2315 7.1072 10.1189
Fading 3.4117 3.8566 5.0211 5.4929 4.4659 2.3007
Total #8.7641 9.7743 9.9014 10.6097 10.6672 11.2714
C.-M. Lin et al.: Guidance Law Evaluation for Missile Guidance Systems 247

8 9

Total = (Glint2 + Target2 + Fading2)1/2 8 Total = (glin2 = Target2 + Fading2)1/2


7
7
6

RMS miss distance (ft)


Glint
Glint 6
RMS miss distance (ft)

5
5 Fading
4 Target 4

3 3 Target
Fading
2 2

1
1
0
0 0 1 2 3 4 5
0 0.5 1 1.5 2 2.5 3 Adjoint Time (second)
Adjoint Time (second)
Fig. 5. RMS miss distance for the PID guidance system (N1 = 3, N2 =
Fig. 3. RMS miss distance for the PN guidance system (N = 3). 1, N3 = 0).

5
were performed with different bandwidths for these models.
4.5 2 2
Total = (glin = Target + Fading ) 2 1/2 The simulation results are listed in Table 5, which shows
4 Glint that the APN guidance system is still better than the PN
3.5 and PID guidance systems, even when different system
RMS miss distance (ft)

parameters are used.


3

2.5
3.3 Optimal time and miss distance sensitivity for tar-
2 get step maneuver
Fading
1.5
Target Other than the discussions in the above, some other
1
sensitivity analyses can be also obtained. Table 6, Table
0.5
7 and Table 8 show the miss distance sensitivities based on
0 a unit step in target lateral acceleration (ft – ft/sec2). From
0 0.5 1 1.5 2 2.5 3
Adjoint Time (second) these tables, it is found that the optimal time for the target
to escape (to maximize the miss distance) before intercep-
Fig. 4. RMS miss distance for the APN guidance system (N = 2). tion is about 0.6049 sec for the PN guidance system with
a navigation constant N = 3, about 1.0207 sec for the APN
bandwidth selection. The simulation results also suggest guidance system with navigation constant N = 6 and
that in general, only the PI guidance law should be used about 0.9029 sec for the PI guidance system with the
instead of the PID guidance law for a missile guidance navigation constants N1 = 2 and N2 = 1. The simulation
system subject to system noises. results are shown in Fig. 6, Fig. 7 and Fig. 8, which show
that if the time of target escape is too early (the adjoint
3.2 RMS miss distance for different system parameters time is large), then the miss distance will be reduced.
These simulation results suggest that pilots of fighter
To widen the scope of study on the effect of the planes, the optimal escape time is about 0.6~1.0 sec before
system parameters in the seeker model, autopilot model, interception. In a well-designed missile guidance system,
and noise filter model shown in Fig. 1, more simulations the miss distance sensitivity curve for a step in target

Table 5. RMS miss distance under difference system parameters.

RMS w3 = 10 w3 = 10 w3 = 10 w3 = 10 w3 = 5 w3 = 5 w3 = 5 w3 = 5
Miss w2 = 10 w2 = 10 w2 = 15 w2 = 15 w2 = 10 w2 = 10 w2 = 15 w2 = 15
Distance (ft) w31 = 20 w1 = 30 w1 = 20 w31 = 30 w1 = 20 w1 = 30 w1 = 20 w1 = 30
PN 7.8301 7.4384 7.5029 7.3334 10.7609 9.5718 8.9875 8.1155
APN 4.7530 4.7695 4.9207 4.9945 5.6207 5.3503 5.2585 5.0578
PI 8.7641 8.1873 8.1230 7.7905 13.8611 12.0102 11.0169 9.6500
248 Asian Journal of Control, Vol. 2, No. 4, December 2000

Table 6. Optimal time and miss distance for the target step maneuver for the PN guidance system.

Navigation Constant N=1 N=2 N=3 N=4 N=5 N=6


optimal time (sec) divergence 1.4985 0.6049 1 0.4600 1.0106 0.8711
&
miss distance (ft) divergence useless #0.0601 0.0401 –0.0339 –0.0389

Table 7. Optimal time and miss distance for the target step maneuver for the APN guidance system.

Navigation Constant N=1 N=2 N=3 N=4 N=5 N=6


optimal time (sec) divergence 0.4705 0.8583 0.6450 1.1795 1.0207
&
miss distance (ft) divergence 0.0192 –0.0320 –0.0399 0.0482 #0.0717

Table 8. Optimal time and miss distance for the target step maneuver for the PI guidance system.

Navigation Constant N1 = 3 N1 = 3 N1 = 4 N1 = 4 N1 = 3 N1 = 2
N2 = 1 N2 = 2 N2 = 1 N2 = 2 N2 = 3 N2 = 1
optimal time (sec) 0.5755 0.5535 0.4432 1.2420 1.6960 0.9029
&
miss distance (ft) 0.0574 0.0551 0.0393 –0.0445 –0.0681 #0.1009

0.07 0.15

0.06
0.1 Optimal time : 0.9029 sec
Miss distance : 0.1009 ft
0.05 Optimal time : 0.6049 sec
Miss distance : 0.0604 ft
Miss distance (ft)
Miss distance (ft)

0.05
0.04

0.03 0

0.02
–0.05
0.01

0 –0.1
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5
Adjoint Time (second) Adjoint Time (second)

Fig. 6. Miss distance due to the target step maneuver for the PN Fig. 8. Miss distance due to the target step maneuver for the PI guidance
guidance system (N = 3). system (N1 = 2, N2 = 1).

0.08
acceleration always approaches zero as the adjoint time
0.06 approaches infinity. The amount of adjoint time required
Optimal time : 1.0207 sec
for this curve to settle down is directly related to the overall
0.04 Miss distance : 0.0717 ft guidance system time constant. Therefore, it can be seen
Miss distance (ft)

that a great deal of information concerning the system


0.02
performance and behavior can be obtained from one
0 adjustment solution.

–0.02
3.4 Miss distance sensitivity for heading error
–0.04
Based on the results of other simulations, the miss
–0.06
0 1 2 3 4 5
distance sensitivities due to the heading error are given in
Adjoint Time (second) Fig. 9 and Fig. 10 for some navigation constants. These
Fig. 7. Miss distance due to the target step maneuver for the APN figures indicate the following:
guidance system (N = 6).
C.-M. Lin et al.: Guidance Law Evaluation for Missile Guidance Systems 249

1
N=1 technique, when applied to a guidance system, obtains not
only the total RMS miss distance, but also the sensitivity
0.5 N=4 function of each factor which contributes to the total RMS
miss distance. Based on analyses, some suggestions for
choosing a suitable guidance law and navigation constant
Miss distance (ft)

0
have been given for missile guidance system design. Also,
N=5 a suggestion for the optimal escape time for pilots of
–0.5
N=2 fighter planes has been given. Besides the guidance laws
N=6
demonstrated in this paper, there exist other guidance
N=3
–1 laws, such as the ideal proportional navigation guidance
law, generalized proportional navigation guidance law,
optimal guidance law, etc. However, these guidance
–1.5
0 1 2 3 4 5 6 systems need information which should be provided by
Adjoint Time (second)
ground equipment, and they can not be formulated as
Fig. 9. Miss distance due to the initial heading error for the PN and APN stand-alone systems in a missile system. Therefore, the
guidance systems (N = 1, 2, …, 6). adjoint technique can not be applied to these systems, so
they have not been included in this paper.
1

N1=3,N2=1
N1=4,N2=2
NOMENCLATURE
0.5 N1=4,N2=1
2

N1=3,N2=2 Us: white noise with power spectral density Φs = B


V
Miss distance (ft)

0 2 4
⋅ ft / sec
Hz
–0.5 N1=2,N2=1
N1=3,N2=3 Usn: white glint noise with power spectral density
Φsn(ft2/Hz)
–1 Ufn: white fading noise with power spectral density
Φfn(rad2/Hz)
B: root mean square target maneuver (ft/sec2)
–1.5
0 1 2 3 4 5 6 τ: average number of crossing zero per second
Adjoint Time (second)
(Hz)
Fig. 10. Miss distance due to the initial heading error for the PI y: relative lateral separation rate (ft/sec)
guidance system y: relative lateral separation (ft)
t f: time of flight (sec)
(1) Excluding the target maneuver, the PN and APN Y(tf): miss distance (ft)
guidance systems produce the same results. The D: seeker dish angle (rad)
settling time required for the miss distance to approach λ: estimated line-of-sight rate (rad/sec)
zero is about 3 sec; however, for the PI guidance n T: target lateral acceleration (ft/sec2)
system, it is about 6 sec. This means that if the time- nc : missile acceleration command (ft/sec2)
to-go for the terminal homing phase is larger than the n L: achieved missile acceleration (ft/sec2)
settling times, the miss distances due to the heading N: navigation constant
error can each be reduced to zero. In addition, the miss Vc: closing velocity
distances due to the heading error can be found in σy(tf): root mean square miss distance (ft)
Fig. 9 and Fig. 10.
(2) The PN guidance system and APN guidance system ACKNOWLEDGEMENT
are more sensitive than the PI guidance system under
different navigation constants. This work was supported by the National Science
Council of the Republic of China under Grant NSC 89-
IV. CONCLUSION 2213-E-155-041.

The best feature of the adjoint technique is that a REFERENCES


great deal of information concerning the system perfor-
mance and behavior can be obtained from “one” run 1. Guelman, M., “Proportional Navigation with a Ma-
adjoint solution for deterministic or stochastic systems; neuvering Target,” IEEE Trans. Aerosp. Electron.
i.e., the adjoint technique only needs one computer run to Syst., Vol. 8, No. 3, pp. 364-371 (1972).
obtain sensitivity function information. The adjoint 2. Willian, F., “Missile Guidance for Low-Altitude Air
250 Asian Journal of Control, Vol. 2, No. 4, December 2000

Defense,” J. Guid. Contr. Dyn., Vol. 2, No. 4, pp. 283- Science Council, during 1998-1999; a Committee Mem-
289 (1979). ber of the Chinese Automatic Control Society in 2000; and
3. Gordon, K. and F. Lee, “Estimation of the Time-to-go the Deputy Chairman of the IEEE Control Systems Society,
Parameter for Air-to-Air Missiles,” J. Guid. Contr. Taipei Section, in 1999. During 1997-1998, he was an
Dyn., Vol. 8, No. 2, pp. 262-266 (1985). honor research fellow at the University of Auckland, New
4. Yang, C.D., F.B. Hsiao and F.B. Yeh, “Generalized Zealand. In 1998 and 1999, he received the Outstanding
Guidance Law for Homing Missiles,” IEEE Trans. Research Award from Yuan Ze University. His research
Aerosp. Electron. Syst., Vol. 25, No. 2, pp. 197-212 interests include control system design, intelligent control
(1989). and systems engineering. He is an IEEE Senior Member,
5. Shukla, U.S. and P.R. Mahapatra, “The Proportional and his name has been listed in Who’s Who in the World.
Navigation Dilemma- Pure or True?” IEEE Trans.
Aerosp. Electron. Syst., Vol. 26, No. 2, pp. 382-392
(1990).
Chun-Fei Hsu was born in Taiwan,
6. Ghose, D., “On the Generlization of True Propor-
Republic of China, in 1974. He re-
tional Navigation,” IEEE Trans. Aerosp. Electron.
ceived the B. S. and M. S. degrees in
Syst., Vol. 30, No. 2, pp. 545-555 (1994).
electrical engineering from Yuan Ze
7. Yuan, P.J. and J.S. Chern, “Exact Closed-Form Solu-
University, Taiwan, Republic of
tion of Generalized Proportional Navigation,” J. Guid.
China in 1997 and 1999, respectively.
Contr. Dyn., Vol. 16, No. 5, pp. 963-966 (1993).
Currently, he is working toward the
8. Yang, C.D. and C.C. Yang, “A Unified Approach to
Ph.D. degree in electrical engineer-
Proportional Navigation,” IEEE Trans. Aerosp.
ing at the same university. His research interests include
Electron. Syst., Vol. 33, No. 2, pp. 557-567 (1997).
motor servo drives, and intelligent control systems includ-
9. Yang, C.D. and C.C. Yang, “Optimal Pure Propor-
ing neural networks and fuzzy logic.
tional Navigation for Maneuvering Targets,” IEEE
Trans. Aerosp. Electron. Syst., Vol. 33, No. 3, pp. 949-
957 (1997).
10. Zarchen, P.,”Complete Statistical Analysis of Nonlin- Shing-Kuo Chang was born in
ear Missile Guidance Systems-SLAM,” J. Guid. Contr. Taiwan, Republic of China, in 1968.
Dyn., Vol. 2, No. 1, pp. 71-78 (1979). He received the M. S. degree in elec-
11. Lin, C.F., Modern Navigation, Guidance, and Con- trical engineering from Yuan Ze
trol Processing, Prentice-Hall, Englewood Cliffs, University, Taiwan, Republic of
New Jersey, pp. 106-111 (1991). China in 1995. He joined the Taiwan
12. Chung, H.Y., C.K. Sun and Y.T. Hsu, “Comparison of Fixed Network Company in 1997.
Laws Among Existing Methods in the Tracking His research interests include flight
Capability,” The 36th Conf. Aeron. Astron., Hsin- control systems, adjoint theory, and systems engineering.
Chu, Taiwan, R.O.C., pp. 639-644 (1994).

Rong-Jong Wai was born in Tainan,


Taiwan, in 1974. He received the B.
Chih-Min Lin was born in Taiwan, S. degree in electrical engineering
Republic of China, in 1959. He re- and the Ph.D. degree in electronic
ceived the B. S. and M. S. degrees in engineering from Chung Yuan Chris-
control engineering and the Ph.D. tian University, Chung-Li, Taiwan,
degree in electronics engineering in 1996 and 1999, respectively. Since
from National Chiao Tung University, 1999, he has been with the Depart-
Taiwan, Republic of China in 1981, ment of Electrical Engineering, Yuan Ze University,
1983 and 1986, respectively. During Chung-Li, Taiwan, where he is currently an Assistant
1986-1992, he was with the Chung Shan Institute of Professor. His research interests include motor servo
Science and Technology as Deputy Director of System drives, mechatronics, nonlinear control, and intelligent
Engineering for a missile system. He also served concur- control systems, including neural networks and fuzzy
rently as an Associate Professor at Chiao Tung University logic. He is a co-author of Intelligent Adaptive Control:
and Chung Yuan University, Taiwan. He joined the Industrial Applications in the Applied Computational In-
faculty of the Department of Electrical Engineering, Yuan telligence Set (CRC Press LLC, 1998) and Drive and
Ze University, Taiwan, in 1993, and since 1997, he has Intelligent Control of Ultrasonic Motor (Taiwan: Tsang-
been a Professor of Electrical Engineering. He also served Hai, 1999). He is a member of the IEEE Industrial
as a Planning Member of the Control Branch, National Electronics Society.

You might also like