基于Q—learning的一种多Agent系统结构模型

多Agent系统是近年来比较热门的一个研究领域，而Q-learning算法是强化学习算法中比较著名的算法，也是应用最广泛的一种强化学习算法。以单Agent强化学习Qlearning算法为基础，提出了一种新的学习协作算法，并根据此算法提出了一种新的多Agent系统体系结构模型，该结构的最大特点是提出了知识共享机制、团队结构思想和引入了服务商概念，最后通过仿真实验说明了该结构体系的优越性。

A Structure Model of Multi-agent System Based on Q-learning

Multi-agent system（MAS） is a very popular research field in recent years, and Q-learning algorithm is a more famous algorithm, also is one of the most widely used reinforcement learning algorithms. In this paper, a new algorithm based on learning cooperation is proposed. Its basis is Q-learning, a single agent reinforcement learning algorithm. Fi nally, a novel structure model of MAS is proposed. The most significant characteristic of the model is to put forward the mechanism of knowledge sharing, the ideal of team structure and to introduce the concept of facilitator. In the end, it shows the advantage of the structure system by the simulation experiment.