The Global Optimization by a Synchronization of Multiple Agents Moving Autonomously with the Chaotic Dynamical Model

カテゴリ: 部門大会

論文No: OS6-4

グループ名: 【C】平成18年電気学会電子・情報・システム部門大会講演論文集

発行日: 2006/09/05

タイトル(英語): The Global Optimization by a Synchronization of Multiple Agents Moving Autonomously with the Chaotic Dynamical Model

著者名: 岡本 卓(Keio University / JSPS Research Fellow [DC2]),相吉英太郎 (Keio University)

著者名(英語): Takashi Okamoto(Keio University / JSPS Research Fellow [DC2]),Eitaro Aiyoshi(Keio University)

キーワード: 大域的最適化|同調現象|カオス|勾配系|マルチエージェントメタヒューリスティクス|Global Optimization|Synchronization Phenomenon|Chaos|Gradient System|Multi-agentMeta-heuristics

要約(日本語): In this study, we propose a new multi-agent type global optimization model using a chaotic dynamical model and a synchronization phenomenon in nonlinear dynamical systems for a continuously differentiable optimization problem. Specifically, firstly, we improve Discrete Gradient Chaos Model (DGCM), which drives each agent's autonomous moving, based on its theoretical analysis. Secondly, a new coupling structure - PD type coupling is derived in order to a stable synchronization of all agents with the chaotic dynamical model in the discrete time system. Finally, we propose a new multi-agent type global optimization model in which each agent autonomously moves driven by improved DGCM and their search trajectories are synchronized to elite agents by PD type coupling model. The proposed model properly achieves the diversification and the intensification which are claimed as important strategies for the global optimization in Meta-heuristics research field. Through applications to proper benchmark problems (in which drawbacks of typical benchmark problems are improved) with 100 variables, we confirm that the proposed model is more effective than other gradient based methods.

PDFファイルサイズ: 6,635 Kバイト