0% found this document useful (0 votes)
108 views

Target Search

Using fuzzy interference

Uploaded by

Anban Sridhar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
108 views

Target Search

Using fuzzy interference

Uploaded by

Anban Sridhar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/251906366

Target search method for a torpedo to the evading ship using fuzzy inference

Conference Paper · January 2009

CITATIONS READS

0 796

6 authors, including:

Jung Min Pak Wooyoung Hong


Wonkwang University Sejong University
32 PUBLICATIONS   493 CITATIONS    21 PUBLICATIONS   57 CITATIONS   

SEE PROFILE SEE PROFILE

Hanseok ko Myo Taeg Lim


University of Maryland, College Park Korea University
277 PUBLICATIONS   1,306 CITATIONS    174 PUBLICATIONS   1,412 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Crowd Density Estimation Using Multi-class Adaboost View project

Optimal FIR filtering that is fundamentally more robust than the Kalman filter. View project

All content following this page was uploaded by Jung Min Pak on 01 October 2014.

The user has requested enhancement of the downloaded file.


ICROS-SICE International Joint Conference 2009
August 18-21, 2009, Fukuoka International Congress Center, Japan

Target Search Method for a Torpedo to the Evading Ship


Using Fuzzy Inference
Jung-Min Pak1, Dong-Gi Woo1, Bonhwa Ku1, Wooyoung Hong2, Hanseok Ko1, Myo-Taeg Lim1
1
School of Electrical Engineering, Korea University, Seoul, Korea
(Tel : +82-2-3290-3243; E-mail: [email protected])
2
Department of Electrical and Electronics Engineering, Korea Naval Academy, Jinhae, Korea
(Tel : +82-55-549-1251; E-mail: [email protected])

Abstract: This Paper proposes target search method for a heavyweight torpedo to detect evading ship effectively by
using fuzzy inference. Warship defends itself by conducting evading maneuver and by using decoy, when a torpedo
attacks it. These days the research on a torpedo is usually focused on how to distinguish a target from decoys. As a result
of the research, a wire-guided torpedo and a wake homing torpedo are developed. In this paper, the efficient target
search method using fuzzy inference against the evasive maneuver of anti-torpedo system is proposed, and proves it by
simulation.

Keywords: Fuzzy Inference, Target Search Method, Heavyweight Torpedo, Evasive Maneuver, Decoy

1. INTRODUCTION self-control heavyweight torpedo. To imitate the


wire-guide torpedo which is controlled by the human,
A torpedo is a typical underwater weapon which is fuzzy inference is used for deciding the searching
launched from a warship, a submarine or an aircraft. direction and the searching range. To evaluate
And it sails under the water by using propeller to attack effectiveness of the proposed method, the Monte-Carlo
an enemy warship. The torpedo is classified by its size method using iterative simulation is used to calculate
as the lightweight torpedo and the heavyweight torpedo. the target hitting probability. Also this paper shows
The torpedo is launched from a surface ship or from an comparative analysis with the existing torpedo logic.
aircraft into water to attack a submarine. A heavyweight Compositions of this thesis are as follow. Chapter 2
torpedo is launched from a submarine to attack a surface shows torpedo combat scenario, motion model of
ship or a submarine. A heavyweight torpedo is big and torpedo and target, acoustic detection model of the
heavy for long distance sailing. The first modern torpedo. In chapter 3, propose the search logic using
torpedo is developed at 1866 by Robert White, England. fuzzy reasoning for the torpedo. Chapter 4 shows
After The World-War II, the torpedo became most evaluation of torpedo effectiveness using proposed
powerful and efficient weapon of the navy, because it method.
shows its ability to destroy an enemy warship at once.
The early torpedoes only can sail straight without a 2. PROBLEM DESCRIPTION AND
guidance system and is launched toward the direction of SCENARIO
target position with high velocity. As the electronic
sensor devices have been developed, modern homing 2.1 Tactical scenario
torpedo with sonar sensor is appeared. The torpedo has
two types homing method. One is a passive homing (1) Surface ship, the target
which detects the sound of target’s noise. The other one In this paper, the destroyer is selected as a target.
is an active homing torpedo which transmits sound Velocity of the destroyer is 18~20 knots when cruising,
waves and detects the reflection of the sound. Against 25~30 knots when performs evasive maneuver. When
the development of homing torpedo, torpedo the destroyer detects an torpedo approaching, it
countermeasures are developed to protect a warship performs evasive maneuver by increasing velocity and
from a torpedo. To neutralize the self-control torpedo changing its direction drastically. Purpose of evasive
which is controlled by its computer brain, tug type maneuver is maximizing distance from the torpedo, so it
decoy and floating decoy are developed. Against the escapes from detection range of the torpedo. When the
decoy, wire-guide torpedo was developed. This kind of warship performs evasive maneuver, evasive course is
the torpedo has a wire attached to the mother ship, so decided according to the bearing of approaching torpedo.
people could control the torpedo. As a result, a torpedo To lure away the torpedo, the destroyer uses
can counteract against the decoys and the jammers, so TACM(Torpedo Acoustic Countermeasure). The tug
accuracy and reliance of the torpedo are increased. type decoy, floating decoy, self-propelled decoy are the
Thereafter, some advanced nations developed the representative TACM. In this paper, the destroyer
advanced torpedo which detects wake created by ship launches four floating type decoys at the same time.
propeller. These days, supercavitation torpedo, which Each decoy is dropped at the position which is decided
sail underwater with super high speed, is being by torpedo alert direction.
researched.
This paper proposes and evaluates the effectiveness
of the new search method using fuzzy inference for a

- 5279 - PR0002/09/0000-5279 ¥400 © 2009 SICE


searching at right side and left side in turn.

Fig. 3 Snake search

In the search mode, if the torpedo succeeds detecting


a target, the torpedo steps into the attack mode and starts
homing guidance. The attack mode is continued until
the torpedo hits a target or loses its target. If the torpedo
cannot find the detected target again in some times, the
torpedo regards it lost its target. In the attack mode, as
the torpedo approaching a target, the torpedo steps into
the final attack mode. At this mode, the speed and the
course steering speed increase so that the torpedo keep
tracking a target and hits a target effectively. If the
torpedo loses its target, it conducts search with re-attack
Fig. 1 Evasive maneuver of a warship search pattern using circle search. If the torpedo loses a
target at the attack mode, the torpedo goes back to the
Looking at the evasion trajectory, when the warship snake search.
changes its course, the direction of the warship cannot
be changed instantly because of its inertia. So the
warship moves straight some distance, and then changes
its course. Motion of the warship is modeled based on
this dynamic characteristic. Fig. 4 Circle search
(2) Torpedo
2.2 Acoustic detection process
When the submarine launches the torpedo, it conducts When a submarine attacks an enemy warship with the
TMA (Target Motion Analysis) using its sensor and it is heavyweight torpedo, noise from driving system is loud.
keeping distance from the target. Using equation (1), The submarine needs to be hidden, so it uses passive
firing angle (launching direction) can be calculated seeker to detect the sound of the warship. The passive
based on target speed, direction, and torpedo speed. And detection is the method that just listens and analyses the
these can be achieved from above analysis. sound signal from a target. To define the successes of
target detection, passive sonar equation (2) is used.
VTE sin φ ASP
Ψ DA = sin −1 ( )
VM (1) SE = SL − TL − NL + DI − DT (2)

Where, SE : Signal Excess


VTE SL : Source Level
TL : Transmission Loss
φ ASP NL : Noise Level
DI : Directivity Index
DT : Detection Threshold
VM
ψ DA After calculating all sonar parameter of the equation
(2), if SE at left side of equation is bigger than 0, then it
is assumed that a target is detected. Meaning of each
sonar parameter and calculation method shows as
below.
VM sin(ψ DA ) = VTE sin(φ ASP ) (1) Source Level (SL)
Fig. 2 Firing angle of torpedo At the passive sonar equation, SL means noise level
of the target. Noise emission of the warship is changed
The launched torpedo steps into the search mode. The according to kind of the warship, cruising speed and
torpedo starts maneuvering with set search pattern. The frequency of sonar. This paper assumes that SL is noise
search pattern is usually the snake search. The snake level of the warship cruising 25 knots which is detected
search is a method which enlarges search area by repeat at 25kHz frequency. Noise of the warship is loudest at

- 5280 -
its tail (Stem, 180°direction) and it is quietest at the Where, Δθ and Δφ are the horizontal and vertical
front(Bow,0°direction). As considering these main lobe width respectively.
characteristices, the source level is calculated by
substract set value from each directions. According to (5) Detection Threshold (TL)
warship noise detecting direction, decreasing value from Finally, DT is set based on maximum search range of
original SL shows at Fig. 5. the torpedo. When the torpedo is placed at maximum
search range from 180° direction (maximum noise
direction of the warship), DT is set to SE becomes zero.

3. SEARCH METHOD

If a target does not use anti-torpedo system, the


torpedo with sonar sensor and homing control system
may have high target hitting rate. But the actual warship
has better sensor than the one torpedo has, so after a
warship detects a torpedo, the warship uses the
anti-torpedo system such as the evasive maneuver and
decoys. This defense system decreases hitting rate of the
torpedo. Therefore the torpedo needs to have logic or a
search method with assuming that a target will use
anti-torpedo method.
The torpedo needs to distinguish noise of the decoy
and noise of the target to counteract the anti-torpedo
system of the warship. And then, it needs to estimate the
Fig. 5 Decrease value of SL according to bearings
evading direction of the target. Noise distinction of the
decoy from the target is belongs to signal processing
(2) Transmission Loss (TL)
area. This paper focuses on how to counteract target’s
TL is calculated through equation (3) with divergence
evasive maneuver, so it assumes that a torpedo can
loss and absorption loss.
recognize the decoy. In the simulation, the torpedo
TL = 20 log R + α × 10−3 recognizes the decoy with level of noise because noise
0.1 f 2 40 f 2 of the decoy very louder than target noise with amount
α= 2
+ 2
+ 0.000275 f 2 + 0.003 (3) of 20dB. If level of the received noise over the threshold
1+ f 4100 + f then it is assumed as a decoy.
Evasive course of the target is estimated by using
In the above equation, the parameter α stands for common principles of evasive maneuver and the
attenuation factor, R is distance between a torpedo and geometrical relation between position of the target and
a target (yard), f is frequency (kHz). the position of the torpedo right after launch. Evasive
course of the warship is decided based on torpedo alert
(3) Noise Level (NL) direction, torpedo speed and speed of itself, but it is
Ambient noise (AN), self noise, reverberation are changed according to the situation. As a result, exact
typical underwater noises. Among these noises, evasive direction is hard to be estimated. But, it is
reverberation is hard to be modeled. So, this paper only obvious that if the torpedo is detected at the right side
considers ambient noise and self noise. The ambient (0°~180°) of a warship then the warship should change
noise can be obtained by using equation (4). its course to the left side (180°~360°). Contrary, if the
warship detects a torpedo at left side then it should
NL = −17.14 log f + 52.643 + 10 log BW (4) change its course to the right side to escape from the
torpedo. Additionally, if the torpedo placed back of the
warship, the probability about the torpedo detects the
where, f stands for frequency(kHz), BW is band width. warship is increased. So the warship should change its
Self noise of the torpedo is added to natural noise at course to avoid such a case. It is because that the
under water. Natural noise is calculated from above propeller at the tail of the warship creates loud noise, so
equation. Self noise is assumed that increased 4dB detection range is longer than noise of front side. With
every 10knot (torpedo speed) these two rules of the warship’s evasion, the estimation
of evasive maneuvering direction of the warship can be
(4) Directivity Index (DI) shown as follows.
DI is calculated with equation (5) which is simplified
(only considering main lobe width). Table 1 Evasive maneuvering direction of a warship
Detected torpedo Front Back
DI = 40 − 10 log(Δθ ⋅ Δφ ) (5) direction
Left Right/Front Right/Back

- 5281 -
Right Left/Front Left/Back

Fig. 6 shows the case that a torpedo approaches to the


warship from right/back direction. In this case, the
warship changes its course to the left/back (180°~270°).

Fig. 7 Membership function of first input, attack angle

Fig. 8 shows the membership functions of detected


decoy direction as the second input variable to the
fuzzy reasoning system. The standard direction is
torpedo advance direction and Left and right side
(between 90°) of the standard is set as Left and Right
each. Center is set at the around of the boundary
between the left and the right.

Fig. 6 Evasive maneuver of a warship in the case that a


torpedo is approaching at the direction of Right/Back

These are outline of the principle of the evasive


maneuver of the warship. Based on these rules, the
torpedo can estimate the evasive course of the warship. Fig. 8 Membership function of second input,
The estimation of the position relation between the Direction of decoy
warship and the torpedo should be qualitative, so
deciding the search direction of a torpedo is hard to be Fig. 9 shows the membership functions of torpedo
made with precise angular calculation. So it is suitable search direction as the fuzzy reasoning output. The
to use qualitative fuzzy rule. In this paper, fuzzy standard direction is the torpedo advance direction. Left
inference is used to estimate the evasive course of the and right of the standard (90°) quantized as LL(Left
target and it is also used to decide the search course for Large), LM(Left Medium), LS(Left Small), St(Straight),
the torpedo which detects an escaping target. RS(Right Small), RM(Right Medium), RL(Right Large).
If the torpedo detects the decoy sound, then it regards Straight should be 0°, but for convenience it is shifted
the target detects the torpedo so it conducts counteract by 90°.
mode instantly. For this action, the torpedo infers best
direction based on the firing angle and the decoy
position. The attack angle involves the information of
moving direction of the torpedo and the target.
Generally the warship tends to increase the distance
from decoy drop position, because if the torpedo is
guided by the decoy, it will increase distance from the
torpedo. Therefore, with the information of an attack
angle and a decoy position, it could be easy to find an
escaping target. This is realized by the fuzzy reasoning.
Fig. 7 is the membership function of the attack angle
which is the first input variable of the fuzzy reasoning Fig. 9 Membership function of output, search course
system. For quantization, the value of attack angle
between 0°~360° are separated 4 sections (Q1, Q2, Q3, Fig. 10 shows a graph of the fuzzy reasoning rule
Q4) and especially the angle of 0° and 90° are set as base. Input variables are the attack angle and the
Front and Rear. detected direction of the decoy. With these inputs, new
search direction which profitable to detect the evading
target is decided as an output of the fuzzy inference. In
this paper, the fuzzy inference uses the Mamdani
method and defuzzification is done by the Centroid

- 5282 -
method. time is getting longer as attack angle goes bigger.
Because torpedo battery has a time limit, short target
hitting time is required. Hitting probability becomes
highest at 90°~120° attack angle. Considering time limit,
140 closer attack angle to 90° is good to launch the torpedo.
120
SearchCourse

Table 3 Hitting probabilities when the warship has no


100
anti-torpedo system
80
Attack
60 Angle 0 30 60 90 120 150 180
40 (deg)
Phit 34 41 34 60 65 57 45
150
100 300 (%)
200
50 109 117 149 214 292 345 346
100 Tave
DecoyDirection 0 0
AttackAngle (sec)
Fig. 10 Surface of fuzzy rule
Fig. 14 shows the target hitting probability with
4. SIMULATION AND ANALYSIS varying TMA information error. It shows that bigger
error has lower target hitting rate.
This chapter evaluates effectiveness of the proposed
search method using fuzzy inference. Target hitting
probability of the torpedo is used to get the measure of
effectiveness (MOE). Hitting target is defined when
torpedo approaches within 50m (half of target length)
from the target center. Simulation method is as follows.
N times errors are generated for TMA information to
calculate torpedo firing angle. Hitting probability is
calculated by hitting time divided by N. Within TMA
information, the target advance direction use -15°~15°
range uniform distribution error, target speed is 3knots
as a Gaussian distribution error. Basic input parameters
of the torpedo and the target for simulation is shown Fig. 11 Comparison of hitting probabilities
below. for three cases TMA errors
Table 2 Values of basic parameters for simulation 4.2 Target conducts evading maneuver
Parameter Name Value The target hitting probability is calculated when a
Target Noise Level 114 dB target warship maneuvers with 20 knots, and increase its
(destroyer, 20knots) speed to 25 knots if it detects the torpedo. The target
Target Speed 20 knots drops 4 decoy to attract the torpedo before conducts
Torpedo Warning Range 2000 m evading maneuver. At this situation, 1) a torpedo has no
Torpedo Speed 35 knots counter search method for decoy, 2) a torpedo can
Sweep Angle 30° distinguish noise of the target and decoy and uses
Receiving Beam Width 30° proposed fuzzy inference method. 3) a torpedo can
Sonar Frequency 25 kHz distinguish noise but has no logic to scan the evading
Max. Detectable Range 1500 m target. With these 3 scenarios, simulation was
conducted.
4.1 Target do not conduct evading maneuver First case, a torpedo has no counteract logic to the
When the warship has no anti-torpedo system, target evading target. As shown table 4, hitting probabilities
hitting probability is calculated as shown at table 3. At are zero with all attack angles.
table 3, hitting probability goes higher as the attack
angle increases. The attack angle represents direction of Table 4 Hitting probabilities when a torpedo has no
the launched torpedo with target advance direction as a counteract logic to the evading target
standard. When a torpedo is launched at in front of the Attack
target, the attack angle is 0°. 180° attack angle is shown Angle 0 30 60 90 120 150 180
when a torpedo is launched at back of the target. As Phit 0 0 0 0 0 0 0
distance is getting closer from the target, target noise is
Tave - - - - - - -
getting louder. As a result, search distance is getting
longer so that target hitting rate is increased with attack
angle goes bigger. But, at table 3, the average hitting Basic logic for a torpedo is to trace most loud noise

- 5283 -
among scanned noise. So, even if all noise wave are 5. CONCLUSION
scanned from the target and decoy, torpedo only trace
the decoy which has 20dB louder noise than the target This paper proposes new search method for a homing
noise. Target use a decoy as a sound curtain to hide heavyweight torpedo to detect the warship which has an
sound its own, so torpedo is hard to scan target noise. anti-torpedo system, and evaluates the effectiveness of
For these reason, it is impossible to hit the target if the method. Among the torpedo countermeasures of the
torpedo has no counter move method. target, the evasive maneuver is the focus of the
Table 5 is shown the result when torpedo is able to proposed method which uses fuzzy inference to estimate
distinguish the target, decoy and has a proposed fuzzy the evading direction and to determine the most
search method. effective search direction. Simulation result shows
average success rate is over 80% when torpedo can
Table 5 Hitting probabilities when a torpedo able to distinguish noise from the target and the decoy.
distinguish the target from the decoy
Attack REFERENCES
Angle 0 30 60 90 120 150 180
Phit 87 52 99 91 100 96 57 [1] A. Mjelde, “A Homing Torpedo : The Effect of the
192 261 348 434 457 471 426 Tactical Situation and the Torpedo Parameters on the
Tave
Torpedo Effectiveness,” Master Thesis, Naval
Postgraduate School, Sep. 1977.
At table 5, hitting probability is over 50% with [2] R. J. Urick, “Principles of Underwater Sound, 3rd
every attack angle, especially over 90% hitting rate is Edition,” McGraw-Hill, 1983.
shown at between 60°~150° attack angle. So, it is [3] A. R. Habayer, “System Effectiveness,” Pergamon
proved that proposed search method using fuzzy is very Press, 1987.
effective. [4] Knut Rief Armo, “The Relationship Between a
One special result is that the average hitting rate is Submarine's Maximum Speed and It's Evasive
about 83% with every attack angle. It is superior to the Ca-pability," Master Thesis, Naval Postgraduate School,
rate when target doesn’t conduct evading motion at table June. 2000.
3. It is because that target hitting rate is increased as [5] Lefteri H. Tsoukalas and Robert E. Uhrig, “Fuzzy
torpedo conducts counter move with fuzzy search and Neural Approaches in Engineering,” John Wiley,
method against every target counter move. 1997.
Proposed method in this paper has more focus on
counter move about the evading target than about the
decoy. So, to prove effectiveness of the proposed
method, one more experiment is required. The result of
the case that a torpedo is able to distinguish the target
and decoy, but has no method for evading target, is
shown at table 6.

Table 6 Hitting probabilities


when a torpedo is able to distinguish the target and
decoy, but has no method for evading target
Attack
Angle 0 30 60 90 120 150 180
Phit 11 100 97 0 0 0 0

Tave 167 225 306 0 0 0 0

At table 6, only with around 0°~30° attack angle


average hitting rate are shown. The other hitting
averages are 0. The reason of hitting rate of 30° and 60°
attack angle are very high, is this simulation assumes
the target conduct evading maneuver only once. When
attack angle is 30°, 60°, evading direction of the target
is going to be parallel to the first advance direction of
the torpedo. So, if the torpedo cut through the decoy and
keep advancing then it can detect the target at once. But,
30° attack angle has lower score than when torpedo has
no fuzzy search method. This is the problem and it
needs additional research.

- 5284 -
View publication stats

You might also like