Towards policies for data insertion in dynamic data driven application systems: a case study sudden ...
Download
1 / 26

Universitat Autònoma de Barcelona, España Roque Rodríguez, Ana Cortés and Tomás Margalef - PowerPoint PPT Presentation


  • 106 Views
  • Uploaded on

Towards policies for data insertion in dynamic data driven application systems: a case study sudden changes in wildland fire. Universitat Autònoma de Barcelona, España Roque Rodríguez, Ana Cortés and Tomás Margalef. Agenda. Problem statement Overview SAPIFE³rt - Real time data injection

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about ' Universitat Autònoma de Barcelona, España Roque Rodríguez, Ana Cortés and Tomás Margalef' - oberon


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

Towards policies for data insertion in dynamic data driven application systems: a case study sudden changes in wildland fire

Universitat Autònoma de Barcelona, España

Roque Rodríguez, Ana Cortés and Tomás Margalef


Agenda
Agenda application systems: a case study sudden changes in wildland fire

  • Problem statement

  • Overview

  • SAPIFE³rt - Real time data injection

  • Policy for data injection

  • Experiments

  • Conclusions and Future Work


Problem statement
Problem application systems: a case study sudden changes in wildland firestatement

  • Forest fires are one of the most worrisome natural disasters, destroying thousands of acres of forests and nearby urban zones, affecting plant, animal and human life.

  • The forest fires are a fact of nature, and have been serving as means of self-regulation of forests. However, these phenomena have become more frequent during the last years.


Problem statement1
Problem statement application systems: a case study sudden changes in wildland fire

  • Fire propagation simulators are a very useful tool to help combat forest fires.

  • Those are based on mathematical and physic models, and with their help, we can mitigate the damage, optimize resources and save lives. But……


Research goals
Research Goals application systems: a case study sudden changes in wildland fire

  • Improve prediction results.

“it is a paradigm whereby application/simulations and measurements become a symbiotic feedback control system. DDDAS entails the ability to dynamically incorporate additional data into an executing application, and in reverse the ability of an application to dynamically steer the measurement process”

Dra. Frederica Darema

  • Reduce execution time.

  • Inject data at execution time.

Applying Dynamic Data Driven Applications Systems concept


Two Stages Propagation Prediction application systems: a case study sudden changes in wildland fire

Simulador

Simulador

Prediction Stage

Parameters

Calibration Stage

Hypothesis: the environmental conditions are similar in the two stages


Calibration stage sapife

Genetic Algorithm application systems: a case study sudden changes in wildland fire

Population

New population

Best

Probability

Selection

Generation

Individual A

Individual B

Bcp1

Bcp2

Acp1

Acp2

Cross

Elitism

Child AB1

Child AB2

Acp1

Bcp2

Bcp1

Acp2

Mutation

scenarios=individuals

Calibration Stage: SAPIFE³

FireSim

Best Population

FireSim

2

2

S

F

M

FireSim

FireSim


Method Evaluations application systems: a case study sudden changes in wildland fire

California Fires

Catalunya Fires

Greece Fires

Real Fires

Prescribed Fires

Synthetic Fires

Error Ratio


Fire Evolution Analysis application systems: a case study sudden changes in wildland fire

Hypothesis: the environmental conditions are similar in the two stages

Fire Spread Evolution

4 to 6 min

Fire Spread Evolution

6 to 8 min


Fire Evolution Analysis application systems: a case study sudden changes in wildland fire

Hypothesis: the environmental conditions are similar in the two stages

Fire Spread Evolution

10 to 12 min

Fire Spread Evolution

12 to 14 min


SAPIFE³ application systems: a case study sudden changes in wildland fire

Real time

Data base

Photo Image

Training Data

Data Stream

Data Collection &

Processing System

Fire Manager

Dynamically Injected Data

Input Parameters

Satellite Image

Simulator

Simulator

Fire Simulated

Weather Station

Genetic

Algorithm

Statistical

Method

Weather Balloon

Urgent HPC


Calibration stage sapife rt

Genetic Algorithm application systems: a case study sudden changes in wildland fire

Population

New population

Probability

Selection

Generation

Individual A

Individual B

Bcp1

Bcp2

Acp1

Acp2

Cross

Elitism

Child AB1

Child AB2

Acp1

Bcp2

Bcp1

Acp2

Mutation

scenarios=individuals

Calibration Stage: SAPIFE³rt

FireSim

Best Population

FireSim

2

2

S

F

M

FireSim

FireSim


Data injection
Data injection application systems: a case study sudden changes in wildland fire

360º

0mph

50º

11

40º

7

45º

9

20mph

WindSpeed bounded range

WindDir bounded range

WindDir 45

WindSpeed 9

Weather Station

Weather Balloon

Best Population

20mph

360º

0mph

7 0.99 0.00 8.00 78.00 0.00 21.00

7 0.99 0.00 4.00 37.00 0.00 21.00

7 0.99 0.00 3.00 45.00 0.00 21.00

WindDir valid range

WindSpeed valid range

7 0.99 0.00 7.00 34.00 0.00 21.00

7 0.99 0.00 5.00 40.00 0.00 21.00

7 0.99 0.00 9.00 38.00 0.00 21.00

7 0.99 0.00 5.00 42.00 0.00 21.00

7 0.99 0.00 4.50 37.00 0.00 21.00

7 0.99 0.00 6.50 39.00 0.00 21.00

7 0.99 0.00 5.00 25.00 0.00 21.00


Data Injection application systems: a case study sudden changes in wildland fire

Map

Map

Map

Map

10

Wind Speed

3

GA

S

GA

S

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

23

9

6

2

24

10

14

20

1

8

15

22

21

3

4

5

11

18

13

7

19

16

25

17

12

P

P

Prediction

Prediction

GA

S


Policy for Data Injection application systems: a case study sudden changes in wildland fire

Change Factor of a given Variable (CFV )

Prediction Stage

Calibration Stage

Prediction Stage

speed

time

c

p

p

c

speed

l

l

l

X

X

X

time

CFV

Changes in the behavior of this variable is negligible

speed

time


CFV Estimation application systems: a case study sudden changes in wildland fire

Map

Map

Map

Map

10

Wind Speed

3

GA

S

GA

S

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

23

9

6

2

24

10

14

20

1

8

15

22

21

3

4

5

11

18

13

7

19

16

25

17

12

P

P

Prediction

Prediction

GA

S


Freeway complex fire
Freeway Complex Fire application systems: a case study sudden changes in wildland fire

  • Injection map every 60 min

  • Injection wind data every 5 min


Results application systems: a case study sudden changes in wildland fire

CFV_threshold= 2.5


Conclusions
Conclusions application systems: a case study sudden changes in wildland fire

  • We observed that data injection in real time can improve the prediction results significantly when conditions are dynamic and changes are sudden.

  • We gain time and flexibility for changing situations.

  • We also conclude that the data acquisition frequency directly affects the prediction results, as well as the precision on the detection of sudden changes.


Applying DDDAS Concept application systems: a case study sudden changes in wildland fire

Feedback

speed

SAPIFE³rt

time

  • Weather Stations

  • Remote Sensing

  • Output

  • Drive Process

  • + or – frequency

  • + or - precision

  • Input Parameters

  • Monitoring

Measurements

speed

time


Thank you
Thank You!!! application systems: a case study sudden changes in wildland fire


Results application systems: a case study sudden changes in wildland fire

  • Injection map every 30 min

  • Injection wind data every 5 min

  • Wind samples data for CFV estimation is 3


Policies for Data Injection application systems: a case study sudden changes in wildland fire

corr=0.97

CFV_threshold=1.5


Results application systems: a case study sudden changes in wildland fire

  • Injection map every 30 min

  • Injection wind data every 5 min

  • Wind samples data for CFV estimation is 3

CFV_threshold= 3.0


Results application systems: a case study sudden changes in wildland fire

  • Injection map every 30 min

  • Injection wind data every 5 min

  • Wind samples data for CFV estimation is 3

CFV_threshold= 3.0


Results application systems: a case study sudden changes in wildland fire

  • Injection map every 60 min

  • Injection wind data every 5 min

  • Wind samples data for CFV estimation is 6


ad