Naslov
AGENT BASED FUZZY COGNITIVE MAPS IN FIRE FIGHTING DECISION SUPPORT

Autori
Štula, Maja ; Stipaničev, Darko ; Vuko, Tomislav

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
14th TIEMS Annual Conference Proceedings / Jones, Alan - Split : Hydrographic Institute of the Republic of Croatia, 2007, 529-537

Skup
TIEMS Annual Conference

Mjesto i datum
Trogir, Hrvatska, 05.06.2007. - 08.06.2007

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Decision support; fire fighting; fuzzy cognitive maps; agent systems; agent based fuzzy

Sažetak
Fire fighting process in Croatia is mainly based on the firefighter knowledge. People on the field decide when to include air forces in the extinction, when to call new forces, etc. Modelling such a complex decision process is difficult and only possible with qualitative modeling techniques. This paper presents the developed agent based fuzzy cognitive map containing fire fighter knowledge. The map can be used as a decision support tool to facilitate the decision process in such a stress situations. Agent based fuzzy cognitive map is qualitative modeling and simulation technique built combining fuzzy cognitive maps and multi-agent systems. Fuzzy Cognitive Maps are qualitative modelling and simulation technique that models system like the group of concepts and the cause-effect relations among concepts. A FCM is represented as a weighted, directed graph with the map concepts as graph nodes and the cause-effect relations as a graph directed edges. FCM is a method for representing the knowledge and an inference process that generates conclusion of the dynamic system behaviour. A multi-agent system can be defined as a network of entities working together on solving the problem that is beyond the agent individual solving capabilities and knowledge with properties like no global system control, decentralized data, etc. The Agent Based Fuzzy Cognitive Map is a Fuzzy Cognitive Map based on a multi-agent system with each concept mapped in to the agent. ABFCM enables each concept in the map to use the different inference process and enables the further extension with the introduction of new agents with specific functionality, like assisting the user with the map results interpretation.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo

POVEZANOST RADA