Study on Multi-agent Simulation System Based on Reinforcement Learning Algorithm
Shu Da Wang, Shuo Ning Wang, Wei Ping Zhang · 2009
Multi-agent simulation system based on reinforcement learning algorithm is a micro-individual acts of modeling and simulation methods, which have wide applicability, distribution, intelligent and interactive features etc. Firstly, studying on reinforcement learning algorithm, and then analysis and design the multi-agent simulation system structure, multi-agent system main modules, the implementation of the definition and finally, carefully design the multi-agent simulation system software, and multi-agent simulation collective system simulation and surrounded the location gathered from the space simulation experiment, the results showed that: Construct a multi-agent simulation system based on reinforcement learning algorithm, achieve real-time simulation of multi-agene, and multi-agent to get effect quickly, and to quickly construct surrounded conduct by mobile groups, the conduct of the system to achieve the global optimum effect.