一种改进的多智能体Q学习算法  

An improve multi-agent system Q-learning algorithm

作　　者：赵辉[1] 赵玉峰[1] 

机构地区：[1]渤海大学工学院,辽宁锦州121013

出　　处：《自动化与仪器仪表》2017年第4期25-27,共3页Automation & Instrumentation

基　　金：国家青年基金项目(61304053)

摘　　要：针对多智能学习特点,提出了一种新的多智能体Q学习算法.算法中将多智能体转换为联合状态的单智能体来学习策略,同时利用改进的随机跳转搜索策略解决了Q算法易陷入局部最优解问题。仿真结果表明,将该算法应用在机械臂轨轨迹划中说明了算法的有效性与泛化能力。Aiming at the study of MAS, we propose an improved MAS Q-learning algorithm, which convert the MAS into single-agent with the combination of the state, a new search strategy is introduced for the problem of local optimal solution in Q- learning. When applied to trajectory planning for manipulator, the simulation results show that the manipulator reaches the target position more quickly and to show the improve Q-learning algorithm is efficient and generalization.

关 键 词：多智能体系统 Q学习 轨迹规划 搜索策略 局部最优 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]