EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
3950. Modelling wind behaviour: using statistical distributions to represent wind speed and direction
Invited abstract in session TA-31: Analytics and the link with stochastic dynamics I, stream Analytics.
Tuesday, 8:30-10:00Room: 046 (building: 208)
Authors (first author is the speaker)
| 1. | Helena Alvelos
|
| DEGEIT / CIDMA, University of Aveiro | |
| 2. | Francisco Marques
|
| Department of Electronics, Telecommunications and Informatics, University of Aveiro | |
| 3. | Ana Raquel Xambre
|
| DEGEIT/CIDMA, University of Aveiro | |
| 4. | Agostinho Agra
|
| Matemática, Universidade de Aveiro | |
| 5. | Filipe Alvelos
|
| Department of Production and Systems, ALGORITMI Research Center / LASI, University of Minho |
Abstract
Wildfires cause significant losses in both material value and human lives. Thus, preventing their occurrence or minimizing their impact when they do happen is paramount. Decision models have thus become increasingly popular in assisting the operational management of firefighting units. The management of wildfires can be divided into: (i) deployment, (ii) dispatch and (iii) positioning. Deployment decisions are made before the fire is detected, dispatch decisions are taken after the ignition is detected, and positioning decisions describe where the dispatched resources should suppress the fire.
The study presented here is part of a project that combines two stages: the prepositioning (deployment) of resources, and the resource movement during the suppression (positioning). Naturally, the problem involves underlying uncertainty, two of its main sources being the ignition location and the wind properties.
The study aims to obtain statistical information about the behaviour of wind speed and wind direction by modelling historical data of a case study region in Portugal. Several statistical distributions were tested using goodness of fit tests, in order to select the most appropriate probability distributions.
The results were then used in the ambiguity set of a mixed integer programming distributionally robust optimization model.
Keywords
- Decision Support Systems
Status: accepted
Back to the list of papers