文献类型：期刊文献

英文题名：Generalized particle model for problem-solving in multi-agent systems

作者：Shuai, Dian-Xun[1,2]; Wang, Xing[1,2]; Feng, Xiang[1,2]

机构：[1] Department of Computer Science and Engineering, East China University of Science and Technology, Shanghai 200237, China; [2] State Key Laboratory of Intelligence Technology and System, Tsinghua University, Beijing 100084, China

年份：2006

卷号：29

期号：5

起止页码：740

外文期刊名：Jisuanji Xuebao/Chinese Journal of Computers

收录：EI(收录号：20062910013932)

语种：英文

外文关键词：Computer simulation - Congestion control (communication) - Dynamics - Kinematics - Mathematical models - Optimization - Parallel algorithms - Problem solving - Stability

摘要：This paper is devoted to a generalized particle model (GPM) approach to distributed problem-solving in MAS, which transforms the optimization problem of resource assignments and task allocations of MAS in complex environment into the kinematics and dynamics in GPM. The complex environment in MAS that the proposed GPM approach may deal with includes: A variety of interactions randomly and concurrently occurring among agents; different personality and autonomy of distinct agents; different life-cycle period, congestion degree and failure rate for distinct entities in MAS. At first, the relation between the GPM and MAS in the context of distributed problem-solving is expatiated. Then the mathematical-physical formalization for GPM and the parallel algorithm GPMA are presented. The basic properties of the GPMA algorithm, including the feasibility, convergency and stability, are discussed. Through a number of simulation experiments and comparisons related to resource assignments and task allocations in MAS in complex environment, the authors demonstrate many advantages of the proposed GPM approach over other coalition methods for MAS problem-solving in terms of the parallelism and the suitability for complex environment.