Experimenting Smarter not Harder - - PDF document

▶
experimenting smarter not harder
SMART_READER_LITE
LIVE PREVIEW

Experimenting Smarter not Harder - - PDF document

Oct 26, 2023 27 likes •128 views

I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Experimenting Smarter not Harder zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA - Taguchi Methods Stella Q"vmd BAE SYSTEMS ISMOR 1 1 1 The Problem Developed

slide-1
SLIDE 1 I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Experimenting Smarter not Harder zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

  • Taguchi Methods
Stella Q"vmd BAE SYSTEMS ISMOR 1 1 1

The Problem

lxmLiw3

Design of Experiments

Developed zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA in Great Brirain zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA in zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA the 1920's Developed extensively i n the academic enyMmeni

1

slide-2
SLIDE 2

I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Taguchi Methods zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Control Factors

RCS Characteristics >-- Force zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Mix

Noise Factors -

Weather zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Threat zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA S A M Density Threat &-Air Density zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

2

slide-3
SLIDE 3 I . zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Configuration Study

6 6 1 1 . Q m n d zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

m m n J

  • B

I

Objectives zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

3

slide-4
SLIDE 4

I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Measures

  • f Effectiveness - 1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

First detections

I

Missiles launched

I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Pk

Failures (Kills)

I

Control Variables Noise Variables

.Scenario .Ingress Route
  • Random seed

4

slide-5
SLIDE 5

l3xz" zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Taguchi Arrays zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

*Balanced Array *Independent Variables Experimental Design Matrix - LTaguchi Array m canpumt)au

I

;= zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

' a

* Numban 1.2 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA I 3 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 8- feplacsd with the Low. Medium or High Value of each tador zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
  • 38
configuratimi out ol the tull fanorial sotol 12.754.y14 mntguratims Uertm,DIL.IOl.m-~m
  • 18 hour8
  • VS. 700 yeam

5

slide-6
SLIDE 6
  • zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
Taguchi Control Factor Separation

1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

1

L l

.........

...

continued

...

I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

I L

. . . . . .

ktr

..... I , . zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

6

slide-7
SLIDE 7

rrrrrrrrmr zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Influence of Control Factors on the three MoEs zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 1 ........ I R zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA *Parameter 15 is most significant *Parameter 3 is not significant for zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA h i t s Control Factors Ranked by Influence on Detections *Parameter 6 is very imponant *Panmeter 2 is unimportant for Detections (cf. Hits &hunches) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA .Panmeter 16 is also relatively unimponant
  • I o D I , - m . ~ - o I I
Control Factor8 Ranked by Influence on Launches *Parameter 6 is less imponant 'Paramem 2 is more imponant .Parameter 16 is very imponant

7

slide-8
SLIDE 8
  • zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
Control Factors Ranked by Influence
  • n Hlts

rEzzmr3

Advantages of using Taguchi

*Reduced Experimentation 36 vs 12,754,584 experiments *Reduction in experimentation time 18 hrs vs over 700 years *Statistically Robust Outcome vs Pseudo Random Sample
  • Structured experimentation
Multiple Analysts reach the same outcome zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

8

slide-9
SLIDE 9

i'

c zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

9

slide-10
SLIDE 10