Abstract: The article deals with the agent-based approach to creating simulation models. A classification is proposed for agent models depending on their behavior and ability to move. Depending on the ability to move and affect other objects, we have been proposed to distinguish between active and passive agents, spatial and time series agents. Particular attention in this article is paid to time series agents and their simulation using neural networks. The article also considers training and testing tool for neural networks developed within the research for the possibility of using neural networks for time series simulation.

Keywords: agent, agent simulation, rescue neural network, time series.

Link:

SIMULATION OF WEATHER IN AGENT MODELS

Vitaliy Lytvynov, Artem Zadorozhnii

http://www.foibg.com/ijima/vol03/ijima03-01-p02.pdf