Multiple Criteria Optimization and Analysis in the

geographertonguesΤεχνίτη Νοημοσύνη και Ρομποτική

30 Νοε 2013 (πριν από 3 χρόνια και 6 μήνες)

63 εμφανίσεις

1

S

ystems

Analysis Laboratory

Helsinki University of Technology

Multiple Criteria Optimization and Analysis in the
Planning of Effects
-
Based Operations (EBO)

Jouni Pousi, Kai Virtanen and Raimo P. Hämäläinen


Systems Analysis Laboratory

Helsinki University of Technology

jouni.pousi@hut.fi, kai.virtanen@hut.fi, raimo@hut.fi

2

S

ystems

Analysis Laboratory

Helsinki University of Technology


Concept for planning and executing military operations

(e.g., Davis, 2001)


Complex military operations, systems perspective


How to produce effects in a system?


Single action produces multiple effects



Effects
-
based operations (EBO)

CONTENTS

Planning of EBO = MCDM problem

Multiple criteria influence diagrams in EBO

3

S

ystems

Analysis Laboratory

Helsinki University of Technology

1.
Identify higher level objective

2.
Describe operation as a system

3.
Derive effects from the

higher
-
level objective


First described qualitatively

4.
Find actions which contribute to the
fulfillment of effects


How to measure the fulfillment

of effects?


Criteria



Steps in EBO planning

Threatening

military buildup

in a country



Public unrest



Etc.




Economic

sanctions



Missile strike



Etc.

Actions

Effects

System

4

S

ystems

Analysis Laboratory

Helsinki University of Technology


Functionally related elements


Elements have states


E.g. works / out of order

Description of the system

Country

Element

Car factory

Element

Steel mill

Dependency

Car factory goes out of business

if steel mill doesn’t produce steel

5

S

ystems

Analysis Laboratory

Helsinki University of Technology


Effects described by one or multiple criteria


Criteria defined in terms of system elements


Multiple elements related to single criterion


Criteria make effects measurable

Qualitative modeling

Criterion

Unemployment

Criterion

Media coverage

Effect

Public

unrest

Country

Car factory

6

S

ystems

Analysis Laboratory

Helsinki University of Technology


System model


Elements = System variables


Dependencies between elements


Actions : Element states


Criteria


The EBO problem





Planning EBO as an MCDM problem

d
x
x
d
x
x
x
x
d

and


feasible
)
,
(
)
(
.
.
))
(
,
),
(
),
(
(
max
2
1
j
j
j
i
i
i
n
g
x
h
x
t
s
f
f
f





)
);
,
(
(
)
(
j
g
f
f
j
j
k
k



x
d
x
)
,
(
j
j
j
g
x


x
d
d
]
,
,
[
1
m
x
x


x
,
)
(
i
i
i
h
x


x
]
,
,
,
,
,
[
1
1
1
m
i
i
i
x
x
x
x






x
7

S

ystems

Analysis Laboratory

Helsinki University of Technology

Planning EBO as an MCDM problem

Criteria

Actions

d
)
(
x
k
f
System

)
(
i
i
i
h
x


x
]
,
,
[
1
m
x
x


x
]
,
,
,
,
[
1
1
1
m
i
i
i
x
x
x
x






x
)
,
(
j
j
j
g
x


x
d


Public unrest



Etc.





Economic


sanctions



Missile strike



Etc.

Actions

Effects

Country


8

S

ystems

Analysis Laboratory

Helsinki University of Technology


Probabilistic modeling
(Davis, 2001)


System dynamics
(Bakken et al., 2004)


Bayesian networks
(Tu et al., 2004)


Single criterion


Combination of Bayesian networks and Petri nets

(Wagenhals & Levis, 2002; Haider & Levis, 2007)


Effects over time


Efficient set not determined


Agent
-
based modeling
(Wallenius & Suzic, 2005)


Calculates criteria given an action


Efficient set not determined


Outranking methods
(Guitouni et al., 2008)


No system model

Previous literature

9

S

ystems

Analysis Laboratory

Helsinki University of Technology


Bayesian network used as a system model


Elements: chance nodes /

random variables


Dependencies: arcs /

conditional probabilities


MCID
(Diehl & Haimes, 2004)


Actions represented by decision nodes


Criteria represented by utility nodes

Multiple criteria influence diagram (MCID)

1
x
6
x
4
x
1
D
3
x
7
x
1
U
m
U
2
x
Criteria

Actions

System

n
D
...

...

5
x
10

S

ystems

Analysis Laboratory

Helsinki University of Technology

EBOLATOR
-

Decision support tool


Implementation utilizing MCID


Construction of system model

(GeNIe, 2009)

11

S

ystems

Analysis Laboratory

Helsinki University of Technology

EBOLATOR
-

Graphical user interface


Visualization of actions


Calculation of efficient set


Criteria weights


Single action

12

S

ystems

Analysis Laboratory

Helsinki University of Technology

EBOLATOR
-

Sensitivity analysis


Weights


MCID probabilities

13

S

ystems

Analysis Laboratory

Helsinki University of Technology

EBOLATOR
-

Example analysis


Defensive air operation


System


Civil and military
infrastructure


Actions


Aircraft positioning and

air combat tactics


MCID


12000 probabilities


729 actions


Analysis


13 efficient actions


Sensitivity analysis

14

S

ystems

Analysis Laboratory

Helsinki University of Technology


Multiple criteria and systems perspective

essential in planning EBO


Similar philosophy applicable in other
application areas (e.g., hospital, marketing)


Previous modeling techniques improved by MCDM


Successful implementation: EBOLATOR


Multiple criteria influence diagram is an interesting
modeling approach in MCDM

Conclusions

15

S

ystems

Analysis Laboratory

Helsinki University of Technology


B. T. Bakken, M. Ruud and S. Johannessen,
“The System Dynamics Approach to
Network Centric Warfare and Effects
-
Based Operations
-

Designing a ``Learning Lab''
for Tomorrow's Military Operations”
, Proceedings of the 22nd International Conference
of the System Dynamics Society, Oxford, England, July 25
-
29, 2004


P. K. Davis,
“Effects
-
Based Operations: A Grand Challenge for the Analytical
Community”
, RAND, 2001


M. Diehl and Y. Y. Haimes,
“Influence Diagram with Multiple Objectives and Tradeoff
Analysis”
, IEEE Transactions on Systems, Man and Cybernetics
-

Part A: Systems and
Humans, vol. 34, no. 3, 2004


A. Guitouni, J. Martel, M. Bélanger and C. Hunter,
“Multiple Criteria Courses of Action
Selection”,
MOR Journal, vol. 13, no. 1, 2008


D
ecision Systems Laboratory of the University of Pittsburgh, “
Graphical Network
Interface”
, http://dsl.sis.pitt.edu, 2009

References 1/2

16

S

ystems

Analysis Laboratory

Helsinki University of Technology


S. Haider and A. H. Levis, ”
Effective Course
-
of
-
Action Determination to Achieve
Desired Effects”,
IEEE Transactions on Systems, Man and Cybernetics
-

Part A:
Systems and Humans, vol. 37, no. 2, 2007


H. Tu, Y. N. Levchuk and K. R. Pattipati,
“Robust Action Strategies to Induce Desired
Effects”
,

IEEE Transactions on Systems, Man and Cybernetics
-

Part A: Systems and
Humans, vol. 34, no. 5, 2004


L. W. Wagenhals and A. H. Levis,
“Modeling Support of Effects
-
Based Operations in
War Games”,
Proceedings of the Command and Control Research and Technology
Symposium, Monterey, California, USA, June 11
-
13, 2002


K. Wallenius and R. Suzic,
“Effects Based Decision Support For Riot Control:
Employing Influence Diagrams and Embedded Simulation”,
Proceedings of the Military
Communications Conference, Atlantic City, New Jersey, USA, October 17
-
20, 2005

References 2/2