マルチエージェントシミュレーションにおけるミクロ的検証手法について(ソフト人工物I)

概要

論文の詳細を見る
There are many social simulation models focused on specific domain. To get the whole vision of the complex intertwined system we should deal with it as a one system that all of the elements might interact with one another. At this point, it is important which scale is concerned. And, several issues will occur owing to a difference of the scale to be governed in the events. These issues have been solved in such way that scale-dependent effects are separated in each event, and a model on each scale is assumed. Therefore, we need model validation method linking between each scale. This paper describes a micro-level validation method of Multi-Agent Simulation (MAS). We focus on a traffic simulation. Clustering spatio-temporal datasets is proposed as the validation method. It is shown that a spatio-temporal data obtained from the movement of each vehicle can be clustered in a space constructed from the summation of absolute values of the accelerations and the average of OV function proposed by Bando et al.
一般社団法人日本機械学会の論文
2004-11-29